Feat/qdrant embedding model discovery (#2086)

Integrated into release/v3.8.0
2026-07-26 09:52:11 +03:00 · 2026-05-10 00:54:04 -03:00
parent a8106bbadd
commit 7d6854e925
47 changed files with 2796 additions and 215 deletions
--- a/Tuto_Qdrant.MD
+++ b/Tuto_Qdrant.MD
@@ -0,0 +1,138 @@
+# Tutorial Qdrant no OmniRoute (Guia para vídeo)
+
+## 1) O que é o Qdrant no OmniRoute
+O Qdrant é o banco vetorial usado para memória semântica.
+
+No OmniRoute, ele ajuda a:
+- Encontrar contexto por significado (não só palavra exata).
+- Reaproveitar memórias antigas com mais precisão.
+- Melhorar respostas com base em histórico relevante.
+- Escalar melhor quando a base de memória cresce.
+
+---
+
+## 2) Quando o OmniRoute envia dados para o Qdrant
+Com Qdrant habilitado e modelo de embedding configurado, o sistema envia vetores quando:
+- Memórias são salvas (upsert de memória).
+- Fluxos de chat recuperam contexto semântico/híbrido.
+- Testes de busca no painel geram embedding e consultam a coleção.
+
+Resumo prático:
+- Sem Qdrant: busca mais limitada (texto/chave).
+- Com Qdrant: busca por similaridade semântica (mais inteligente).
+
+---
+
+## 3) Pré-requisitos
+Você precisa de:
+- Instância Qdrant acessível (porta 6333).
+- Coleção criada (ex.: `omniroute_memory`).
+- Modelo de embedding válido (ex.: OpenRouter).
+- Credencial do provider do embedding configurada no OmniRoute.
+
+Exemplo de modelo OpenRouter:
+- `openrouter/nvidia/llama-nemotron-embed-v1-1b-v2:free`
+
+Importante:
+- O texto do modelo deve estar em formato `provider/model`.
+- Se usar modelo com dimensão diferente da coleção, a busca falha.
+
+---
+
+## 4) Como configurar no painel do OmniRoute
+No menu:
+- `Admin > Settings > Qdrant (Memória vetorial)`
+
+Preencha:
+- `Ativar Qdrant`: ligado.
+- `Host`: IP ou URL do servidor Qdrant (sem porta no campo Host).
+- `Porta`: `6333`.
+- `Collection`: `omniroute_memory` (ou nome que você criou).
+- `Modelo de embedding`: selecione da lista ou digite manualmente.
+- `API Key`: opcional (preencha se seu Qdrant exigir).
+
+Depois:
+1. Clique em `Salvar`.
+2. Clique em `Testar conexão`.
+3. No `Teste de busca`, digite um texto e clique em `Buscar`.
+
+---
+
+## 5) Como criar a coleção no Dashboard do Qdrant (sem comando)
+No Qdrant Dashboard:
+1. Clique em `Create collection`.
+2. Escolha `Global search`.
+3. Em tipo de busca, use `Custom`.
+4. Configure vetor:
+   - Vector name: `omniao` (padrão esperado pelo OmniRoute atualmente).
+   - Size: dimensão do seu modelo de embedding (ex.: 2048 em alguns modelos NVIDIA).
+   - Distance: `Cosine`.
+5. Salve a coleção com nome `omniroute_memory`.
+
+Se já tinha coleção com dimensão errada:
+- Recrie a coleção com dimensão correta.
+
+---
+
+## 6) Como validar se está funcionando
+Checklist rápido:
+1. `Testar conexão` no OmniRoute retorna OK.
+2. Busca no painel retorna resultados (não “Sem resultados”).
+3. No Qdrant Dashboard, aparecem pontos na coleção (payload + vector).
+4. Resultados de chat passam a recuperar contexto mais relevante.
+
+Sinal clássico de problema:
+- Dados entram no Qdrant, mas busca do painel não retorna nada.
+
+Causas comuns:
+- Dimensão do vetor incompatível.
+- Nome do vetor diferente do esperado (`omniao`).
+- Modelo inválido/incompleto no campo de embedding.
+- Provider sem credencial ativa.
+
+---
+
+## 7) O que melhorou com esta atualização
+Nesta melhoria do OmniRoute:
+- Suporte a embeddings de qualquer provider compatível (não só OpenAI fixo).
+- Endpoint para carregar modelos de embedding na tela de configurações.
+- Campo manual para modelo custom quando não aparecer na lista.
+- Ajuda visual (`?`) com passo rápido de configuração Qdrant + OpenRouter.
+
+---
+
+## 8) Roteiro curto para seu vídeo
+Sugestão de demo (3-5 minutos):
+1. Mostrar problema sem Qdrant (busca simples).
+2. Abrir Settings e habilitar Qdrant.
+3. Configurar host/porta/collection/modelo.
+4. Salvar + testar conexão.
+5. Fazer `Teste de busca` no painel.
+6. Abrir Qdrant Dashboard e mostrar ponto salvo + vetor.
+7. Rodar um chat e mostrar melhoria de recuperação semântica.
+
+Mensagem final para a galera:
+- "Qdrant no OmniRoute transforma memória de palavra-chave em memória por significado."
+
+---
+
+## 9) Referências de código (para equipe técnica)
+- UI de configuração Qdrant:
+  - `src/app/(dashboard)/dashboard/settings/components/MemorySkillsTab.tsx`
+- Endpoint de modelos de embedding para Qdrant:
+  - `src/app/api/settings/qdrant/embedding-models/route.ts`
+- Integração backend com Qdrant (health, upsert, search, cleanup):
+  - `src/lib/memory/qdrant.ts`
+- Recuperação de memórias no fluxo de chat:
+  - `src/lib/memory/retrieval.ts`
+  - `open-sse/handlers/chatCore.ts`
+
+---
+
+## 10) Observação importante de segurança
+Nunca exponha em vídeo:
+- API key completa do OpenRouter.
+- Tokens reais de produção.
+- Endpoints internos sem proteção.
+
+Use chaves mascaradas e ambiente de demonstração.
--- a/src/app/(dashboard)/dashboard/settings/components/MemorySkillsTab.tsx
+++ b/src/app/(dashboard)/dashboard/settings/components/MemorySkillsTab.tsx
@@ -13,6 +13,21 @@ interface MemoryConfig {
  skillsEnabled: boolean;
 }

+interface QdrantSettings {
+  enabled: boolean;
+  host: string;
+  port: number;
+  collection: string;
+  embeddingModel: string;
+  hasApiKey: boolean;
+  apiKeyMasked: string | null;
+}
+
+interface EmbeddingModelOption {
+  value: string;
+  label: string;
+}
+
 const STRATEGIES = [
  { value: "recent", labelKey: "recent", descKey: "recentDesc" },
  { value: "semantic", labelKey: "semantic", descKey: "semanticDesc" },
@@ -30,6 +45,35 @@ export default function MemorySkillsTab() {
  const [loading, setLoading] = useState(true);
  const [saving, setSaving] = useState(false);
  const [status, setStatus] = useState("");
+
+  const [qdrant, setQdrant] = useState<QdrantSettings>({
+    enabled: false,
+    host: "",
+    port: 6333,
+    collection: "omniroute_memory",
+    embeddingModel: "openai/text-embedding-3-small",
+    hasApiKey: false,
+    apiKeyMasked: null,
+  });
+  const [qdrantApiKeyInput, setQdrantApiKeyInput] = useState("");
+  const [qdrantSaving, setQdrantSaving] = useState(false);
+  const [qdrantStatus, setQdrantStatus] = useState<"" | "saved" | "error">("");
+  const [qdrantHealth, setQdrantHealth] = useState<{
+    ok: boolean;
+    latencyMs: number;
+    error?: string;
+  } | null>(null);
+  const [qdrantChecking, setQdrantChecking] = useState(false);
+  const [qdrantQuery, setQdrantQuery] = useState("");
+  const [qdrantSearching, setQdrantSearching] = useState(false);
+  const [qdrantResults, setQdrantResults] = useState<
+    Array<{ id: string; score: number; payload?: Record<string, unknown> }>
+  >([]);
+  const [qdrantCleanupLoading, setQdrantCleanupLoading] = useState(false);
+  const [qdrantCleanupMsg, setQdrantCleanupMsg] = useState("");
+  const [embeddingOptions, setEmbeddingOptions] = useState<EmbeddingModelOption[]>([]);
+  const [qdrantHelpOpen, setQdrantHelpOpen] = useState(false);
+
  const [skillsmpApiKey, setSkillsmpApiKey] = useState("");
  const [skillsmpSaving, setSkillsmpSaving] = useState(false);
  const [skillsmpStatus, setSkillsmpStatus] = useState("");
@@ -42,12 +86,21 @@ export default function MemorySkillsTab() {
    Promise.all([
      fetch("/api/settings/memory").then((res) => (res.ok ? res.json() : null)),
      fetch("/api/settings").then((res) => (res.ok ? res.json() : null)),
+      fetch("/api/settings/qdrant").then((res) => (res.ok ? res.json() : null)),
+      fetch("/api/settings/qdrant/embedding-models").then((res) => (res.ok ? res.json() : null)),
    ])
-      .then(([memData, settingsData]) => {
+      .then(([memData, settingsData, qdrantData, embeddingData]) => {
        if (memData) setConfig(memData);
        if (settingsData?.skillsmpApiKey) {
          setSkillsmpApiKey(settingsData.skillsmpApiKey);
        }
+        if (qdrantData) {
+          setQdrant(qdrantData);
+          setQdrantApiKeyInput("");
+        }
+        if (embeddingData?.models && Array.isArray(embeddingData.models)) {
+          setEmbeddingOptions(embeddingData.models);
+        }
        if (
          settingsData?.skillsProvider === "skillsmp" ||
          settingsData?.skillsProvider === "skillssh"
@@ -56,7 +109,111 @@ export default function MemorySkillsTab() {
        }
      })
      .catch(() => {})
-      .finally(() => setLoading(false));
+      .finally(() => {
+        setLoading(false);
+      });
+  }, []);
+
+  const saveQdrant = useCallback(
+    async (updates: Partial<QdrantSettings> & { apiKey?: string }) => {
+      const previous = qdrant;
+      const next = { ...qdrant, ...updates };
+      setQdrant(next);
+      setQdrantSaving(true);
+      setQdrantStatus("");
+      try {
+        const res = await fetch("/api/settings/qdrant", {
+          method: "PUT",
+          headers: { "Content-Type": "application/json" },
+          body: JSON.stringify({
+            enabled: next.enabled,
+            host: next.host,
+            port: next.port,
+            collection: next.collection,
+            embeddingModel: next.embeddingModel,
+            ...(updates.apiKey !== undefined ? { apiKey: updates.apiKey } : {}),
+          }),
+        });
+        if (res.ok) {
+          const data = await res.json().catch(() => next);
+          setQdrant(data);
+          setQdrantApiKeyInput("");
+          setQdrantStatus("saved");
+          setTimeout(() => setQdrantStatus(""), 2000);
+        } else {
+          setQdrant(previous);
+          setQdrantStatus("error");
+        }
+      } catch {
+        setQdrant(previous);
+        setQdrantStatus("error");
+      } finally {
+        setQdrantSaving(false);
+      }
+    },
+    [qdrant]
+  );
+
+  const checkQdrant = useCallback(async () => {
+    setQdrantChecking(true);
+    try {
+      const res = await fetch("/api/settings/qdrant/health");
+      if (res.ok) setQdrantHealth(await res.json());
+      else setQdrantHealth({ ok: false, latencyMs: 0, error: "HTTP error" });
+    } catch (e) {
+      setQdrantHealth({
+        ok: false,
+        latencyMs: 0,
+        error: e instanceof Error ? e.message : String(e),
+      });
+    } finally {
+      setQdrantChecking(false);
+    }
+  }, []);
+
+  const testQdrantSearch = useCallback(async () => {
+    const q = qdrantQuery.trim();
+    if (!q) return;
+    setQdrantSearching(true);
+    setQdrantResults([]);
+    try {
+      const res = await fetch("/api/settings/qdrant/search", {
+        method: "POST",
+        headers: { "Content-Type": "application/json" },
+        body: JSON.stringify({ query: q, topK: 5 }),
+      });
+      const data = await res.json().catch(() => null);
+      if (res.ok && data?.ok) {
+        setQdrantResults(Array.isArray(data.results) ? data.results : []);
+      } else {
+        setQdrantResults([]);
+      }
+    } catch {
+      setQdrantResults([]);
+    } finally {
+      setQdrantSearching(false);
+    }
+  }, [qdrantQuery]);
+
+  const runQdrantCleanup = useCallback(async () => {
+    setQdrantCleanupLoading(true);
+    setQdrantCleanupMsg("");
+    try {
+      const res = await fetch("/api/settings/qdrant/cleanup", { method: "POST" });
+      const data = await res.json().catch(() => null);
+      if (res.ok && data?.ok) {
+        setQdrantCleanupMsg(
+          `OK: removeu ${data.deletedCount ?? 0} ponto(s) (retencao: ${data.retentionDays} dias)`
+        );
+      } else {
+        const err = data?.error || "Falha na limpeza";
+        setQdrantCleanupMsg(`Erro: ${String(err)}`);
+      }
+    } catch (e) {
+      setQdrantCleanupMsg(`Erro: ${e instanceof Error ? e.message : String(e)}`);
+    } finally {
+      setQdrantCleanupLoading(false);
+    }
  }, []);

  const saveSkillsmpApiKey = useCallback(async () => {
@@ -280,6 +437,280 @@ export default function MemorySkillsTab() {
        )}
      </Card>

+      {/* Qdrant (optional semantic memory index) */}
+      <Card data-testid="qdrant-settings-card">
+        <div className="flex items-center gap-3 mb-5">
+          <div className="p-2 rounded-lg bg-emerald-500/10 text-emerald-500">
+            <span className="material-symbols-outlined text-[20px]" aria-hidden="true">
+              database
+            </span>
+          </div>
+          <div>
+            <h3 className="text-lg font-semibold">{t("qdrantTitle")}</h3>
+            <p className="text-sm text-text-muted">{t("qdrantDesc")}</p>
+          </div>
+
+          <span
+            className={`ml-auto inline-flex items-center gap-2 text-xs font-medium ${
+              qdrant.enabled
+                ? qdrantHealth?.ok
+                  ? "text-emerald-500"
+                  : "text-red-500"
+                : "text-text-muted"
+            }`}
+          >
+            <span
+              className={`inline-block w-2.5 h-2.5 rounded-full ${
+                qdrant.enabled ? (qdrantHealth?.ok ? "bg-emerald-500" : "bg-red-500") : "bg-border"
+              }`}
+              aria-hidden="true"
+            />
+            {qdrant.enabled
+              ? qdrantHealth?.ok
+                ? t("qdrantStatusActive")
+                : t("qdrantStatusError")
+              : t("qdrantStatusDisabled")}
+          </span>
+        </div>
+
+        <div className="flex items-center justify-between p-4 rounded-lg bg-surface/30 border border-border/30 mb-4">
+          <div>
+            <p className="text-sm font-medium">{t("qdrantEnable")}</p>
+            <p className="text-xs text-text-muted mt-0.5">{t("qdrantEnableDesc")}</p>
+          </div>
+          <div className="flex items-center gap-2">
+            <button
+              onClick={checkQdrant}
+              disabled={qdrantChecking || qdrantSaving}
+              className="px-3 h-8 text-xs font-medium rounded-lg bg-white/5 border border-border/60 hover:bg-white/10 disabled:opacity-50 transition-colors"
+            >
+              {qdrantChecking ? t("qdrantTesting") : t("qdrantTestConnection")}
+            </button>
+            <button
+              data-testid="qdrant-enabled-switch"
+              onClick={() => saveQdrant({ enabled: !qdrant.enabled })}
+              disabled={qdrantSaving}
+              className={`relative w-11 h-6 rounded-full transition-colors ${
+                qdrant.enabled ? "bg-emerald-500" : "bg-border"
+              }`}
+              role="switch"
+              aria-checked={qdrant.enabled}
+            >
+              <span
+                className={`absolute top-1 left-1 w-4 h-4 bg-white rounded-full transition-transform ${
+                  qdrant.enabled ? "translate-x-5" : "translate-x-0"
+                }`}
+              />
+            </button>
+          </div>
+        </div>
+
+        {qdrantStatus === "saved" && (
+          <div className="mb-4 text-xs font-medium text-emerald-500 flex items-center gap-1">
+            <span className="material-symbols-outlined text-[14px]">check_circle</span>{" "}
+            {t("qdrantSaved")}
+          </div>
+        )}
+        {qdrantStatus === "error" && (
+          <div className="mb-4 text-xs font-medium text-red-500">{t("qdrantSaveError")}</div>
+        )}
+
+        <div className="grid grid-cols-1 md:grid-cols-2 gap-3">
+          <div className="p-4 rounded-lg bg-surface/30 border border-border/30">
+            <label className="text-sm font-medium block mb-2">Host</label>
+            <input
+              value={qdrant.host}
+              onChange={(e) => setQdrant((s) => ({ ...s, host: e.target.value }))}
+              placeholder="http://127.0.0.1"
+              className="w-full px-3 py-2 rounded-lg bg-background border border-border text-sm font-mono focus:outline-none focus:ring-1 focus:ring-emerald-500"
+            />
+            <p className="text-xs text-text-muted mt-2">{t("qdrantHostHint")}</p>
+          </div>
+
+          <div className="p-4 rounded-lg bg-surface/30 border border-border/30">
+            <label className="text-sm font-medium block mb-2">{t("qdrantPort")}</label>
+            <input
+              value={qdrant.port}
+              onChange={(e) =>
+                setQdrant((s) => ({
+                  ...s,
+                  port: Math.max(1, Math.min(65535, Number(e.target.value) || 0)),
+                }))
+              }
+              placeholder="6333"
+              className="w-full px-3 py-2 rounded-lg bg-background border border-border text-sm font-mono focus:outline-none focus:ring-1 focus:ring-emerald-500"
+            />
+            <p className="text-xs text-text-muted mt-2">{t("qdrantPortHint")}</p>
+          </div>
+
+          <div className="p-4 rounded-lg bg-surface/30 border border-border/30">
+            <label className="text-sm font-medium block mb-2">Collection</label>
+            <input
+              value={qdrant.collection}
+              onChange={(e) => setQdrant((s) => ({ ...s, collection: e.target.value }))}
+              placeholder="omniroute_memory"
+              className="w-full px-3 py-2 rounded-lg bg-background border border-border text-sm font-mono focus:outline-none focus:ring-1 focus:ring-emerald-500"
+            />
+            <p className="text-xs text-text-muted mt-2">{t("qdrantCollectionHint")}</p>
+          </div>
+
+          <div className="p-4 rounded-lg bg-surface/30 border border-border/30">
+            <div className="flex items-center gap-2 mb-2">
+              <label className="text-sm font-medium block">{t("qdrantEmbeddingModel")}</label>
+              <button
+                type="button"
+                onClick={() => setQdrantHelpOpen((v) => !v)}
+                className="inline-flex items-center justify-center w-5 h-5 rounded-full border border-border/70 text-xs text-text-muted hover:bg-white/10"
+                title={t("qdrantHelpTitle")}
+                aria-label={t("qdrantHelpTitle")}
+              >
+                ?
+              </button>
+            </div>
+            {qdrantHelpOpen && (
+              <div className="mb-3 p-3 rounded-lg bg-background/60 border border-border/60 text-xs text-text-muted leading-relaxed">
+                <p className="font-medium text-white mb-1">{t("qdrantHelpQuickTitle")}</p>
+                <p>{t("qdrantHelpStep1")}</p>
+                <p>{t("qdrantHelpStep2")}</p>
+                <p>{t("qdrantHelpStep3")}</p>
+                <p>{t("qdrantHelpStep4")}</p>
+              </div>
+            )}
+            <select
+              value=""
+              onChange={(e) => {
+                const value = e.target.value;
+                if (value) setQdrant((s) => ({ ...s, embeddingModel: value }));
+              }}
+              className="w-full px-3 py-2 rounded-lg bg-background border border-border text-sm mb-2 focus:outline-none focus:ring-1 focus:ring-emerald-500"
+            >
+              <option value="">{t("qdrantEmbeddingQuickSelect")}</option>
+              {embeddingOptions.map((opt) => (
+                <option key={opt.value} value={opt.value}>
+                  {opt.value}
+                </option>
+              ))}
+            </select>
+            <input
+              value={qdrant.embeddingModel}
+              onChange={(e) => setQdrant((s) => ({ ...s, embeddingModel: e.target.value }))}
+              placeholder={t("qdrantEmbeddingInputPlaceholder")}
+              className="w-full px-3 py-2 rounded-lg bg-background border border-border text-sm font-mono focus:outline-none focus:ring-1 focus:ring-emerald-500"
+            />
+            <p className="text-xs text-text-muted mt-2">{t("qdrantEmbeddingHint")}</p>
+          </div>
+
+          <div className="p-4 rounded-lg bg-surface/30 border border-border/30 md:col-span-2">
+            <label className="text-sm font-medium block mb-2">
+              API Key ({t("optional")}){" "}
+              {qdrant.hasApiKey && qdrant.apiKeyMasked ? (
+                <span className="text-xs text-text-muted font-mono">
+                  {t("current")}: {qdrant.apiKeyMasked}
+                </span>
+              ) : null}
+            </label>
+            <div className="flex gap-2">
+              <input
+                type="password"
+                value={qdrantApiKeyInput}
+                onChange={(e) => setQdrantApiKeyInput(e.target.value)}
+                placeholder={
+                  qdrant.hasApiKey
+                    ? t("qdrantApiKeyPlaceholderKeep")
+                    : t("qdrantApiKeyPlaceholderOptional")
+                }
+                className="flex-1 px-3 py-2 rounded-lg bg-background border border-border text-sm font-mono focus:outline-none focus:ring-1 focus:ring-emerald-500"
+              />
+              {qdrant.hasApiKey && (
+                <button
+                  onClick={() => saveQdrant({ apiKey: "" })}
+                  disabled={qdrantSaving}
+                  className="px-3 py-2 text-sm font-medium rounded-lg bg-white/5 border border-border/60 hover:bg-white/10 disabled:opacity-50 transition-colors"
+                >
+                  {t("remove")}
+                </button>
+              )}
+              <button
+                onClick={() =>
+                  saveQdrant(
+                    qdrantApiKeyInput.trim().length > 0 ? { apiKey: qdrantApiKeyInput } : {}
+                  )
+                }
+                disabled={qdrantSaving}
+                className="px-4 py-2 text-sm font-medium rounded-lg bg-emerald-500 text-white hover:bg-emerald-600 disabled:opacity-50 transition-colors"
+              >
+                {qdrantSaving ? t("saving") : t("save")}
+              </button>
+            </div>
+            <p className="text-xs text-text-muted mt-2">{t("qdrantSaveHint")}</p>
+          </div>
+        </div>
+
+        <div className="mt-4 p-4 rounded-lg bg-surface/30 border border-border/30">
+          <div className="flex items-center justify-between gap-2">
+            <div>
+              <p className="text-sm font-medium">{t("qdrantSearchTestTitle")}</p>
+              <p className="text-xs text-text-muted mt-0.5">{t("qdrantSearchTestDesc")}</p>
+            </div>
+            <button
+              onClick={testQdrantSearch}
+              disabled={qdrantSearching}
+              className="px-4 py-2 text-sm font-medium rounded-lg bg-white/5 border border-border/60 hover:bg-white/10 disabled:opacity-50 transition-colors"
+            >
+              {qdrantSearching ? t("searching") : t("search")}
+            </button>
+          </div>
+          <div className="mt-3 flex gap-2">
+            <input
+              value={qdrantQuery}
+              onChange={(e) => setQdrantQuery(e.target.value)}
+              placeholder={t("qdrantSearchPlaceholder")}
+              className="flex-1 px-3 py-2 rounded-lg bg-background border border-border text-sm focus:outline-none focus:ring-1 focus:ring-emerald-500"
+            />
+          </div>
+          {qdrantResults.length > 0 && (
+            <div className="mt-3 space-y-2">
+              {qdrantResults.map((r) => (
+                <div key={r.id} className="p-3 rounded-lg bg-background/40 border border-border/40">
+                  <div className="flex items-center justify-between">
+                    <span className="text-xs font-mono text-text-muted">{r.id}</span>
+                    <span className="text-xs font-mono text-emerald-400">
+                      score {r.score.toFixed(4)}
+                    </span>
+                  </div>
+                  <div className="mt-2 text-xs text-text-muted">
+                    {(r.payload?.key as string) ? `key: ${String(r.payload?.key)}` : null}
+                  </div>
+                </div>
+              ))}
+            </div>
+          )}
+          {qdrantResults.length === 0 && qdrantQuery.trim().length > 0 && !qdrantSearching && (
+            <p className="mt-3 text-xs text-text-muted">{t("qdrantNoResults")}</p>
+          )}
+        </div>
+
+        <div className="mt-4 p-4 rounded-lg bg-surface/30 border border-border/30">
+          <div className="flex items-center justify-between gap-2">
+            <div>
+              <p className="text-sm font-medium">{t("qdrantCleanupTitle")}</p>
+              <p className="text-xs text-text-muted mt-0.5">
+                {t("qdrantCleanupDesc")} {t("retentionDays")} ({config.retentionDays} {t("days")}).
+              </p>
+            </div>
+            <button
+              onClick={runQdrantCleanup}
+              disabled={qdrantCleanupLoading}
+              className="px-4 py-2 text-sm font-medium rounded-lg bg-white/5 border border-border/60 hover:bg-white/10 disabled:opacity-50 transition-colors"
+            >
+              {qdrantCleanupLoading ? t("cleaning") : t("cleanNow")}
+            </button>
+          </div>
+          {qdrantCleanupMsg && <p className="mt-2 text-xs text-text-muted">{qdrantCleanupMsg}</p>}
+        </div>
+      </Card>
+
+      {/* Skills Settings (placeholder) */}
      <Card data-testid="skills-settings-card">
        <div className="flex items-center gap-3 mb-5">
          <div className="p-2 rounded-lg bg-amber-500/10 text-amber-500">
--- a/src/app/api/settings/qdrant/embedding-models/route.ts
+++ b/src/app/api/settings/qdrant/embedding-models/route.ts
@@ -0,0 +1,93 @@
+import { NextRequest, NextResponse } from "next/server";
+import { isAuthenticated } from "@/shared/utils/apiAuth";
+import { AI_MODELS } from "@/shared/constants/models";
+import { getProviderConnections } from "@/lib/db/providers";
+
+type EmbeddingModelOption = {
+  value: string;
+  label: string;
+};
+
+function isLikelyEmbeddingModel(provider: string, model: string, name: string): boolean {
+  const haystack = `${provider}/${model} ${name}`.toLowerCase();
+  if (haystack.includes("embedding")) return true;
+  if (haystack.includes("embed")) return true;
+  if (haystack.includes("text-embedding")) return true;
+  return false;
+}
+
+export async function GET(request: NextRequest) {
+  if (!(await isAuthenticated(request))) {
+    return NextResponse.json({ error: "Unauthorized" }, { status: 401 });
+  }
+
+  try {
+    const options: EmbeddingModelOption[] = AI_MODELS.filter((m: any) =>
+      isLikelyEmbeddingModel(String(m.provider || ""), String(m.model || ""), String(m.name || ""))
+    )
+      .map((m: any) => ({
+        value: `${m.provider}/${m.model}`,
+        label: `${m.provider}/${m.model} - ${m.name}`,
+      }))
+      .sort((a, b) => a.value.localeCompare(b.value));
+
+    // Add OpenRouter account models that explicitly support embeddings.
+    try {
+      const connections = (await getProviderConnections({
+        provider: "openrouter",
+        isActive: true,
+      })) as Array<Record<string, unknown>>;
+      const apiKey = connections.find(
+        (c) => typeof c.apiKey === "string" && (c.apiKey as string).trim().length > 0
+      )?.apiKey as string | undefined;
+
+      if (apiKey) {
+        const controller = new AbortController();
+        const timeout = setTimeout(() => controller.abort(), 7000);
+        let res: Response;
+        try {
+          res = await fetch("https://openrouter.ai/api/v1/models?output_modalities=embeddings", {
+            method: "GET",
+            headers: {
+              Authorization: `Bearer ${apiKey}`,
+            },
+            cache: "no-store",
+            signal: controller.signal,
+          });
+        } finally {
+          clearTimeout(timeout);
+        }
+        if (res.ok) {
+          const data = (await res.json().catch(() => null)) as any;
+          const rows = Array.isArray(data?.data) ? data.data : [];
+          for (const row of rows) {
+            const id = typeof row?.id === "string" ? row.id.trim() : "";
+            if (!id) continue;
+            const value = `openrouter/${id}`;
+            if (options.some((o) => o.value === value)) continue;
+            options.push({
+              value,
+              label: `${value} - ${String(row?.name || id)}`,
+            });
+          }
+        }
+      }
+    } catch {
+      // Best effort only: keep endpoint fast and resilient.
+    }
+
+    // Ensure the default always exists as a safe fallback.
+    if (!options.some((o) => o.value === "openai/text-embedding-3-small")) {
+      options.unshift({
+        value: "openai/text-embedding-3-small",
+        label: "openai/text-embedding-3-small - OpenAI Text Embedding 3 Small",
+      });
+    }
+
+    options.sort((a, b) => a.value.localeCompare(b.value));
+
+    return NextResponse.json({ models: options });
+  } catch (error) {
+    return NextResponse.json({ error: String(error), models: [] }, { status: 500 });
+  }
+}
--- a/src/app/api/v1/embeddings/route.ts
+++ b/src/app/api/v1/embeddings/route.ts
@@ -1,35 +1,22 @@
-import { handleEmbedding } from "@omniroute/open-sse/handlers/embeddings.ts";
-import {
-  getProviderCredentials,
-  clearRecoveredProviderState,
-  extractApiKey,
-  isValidApiKey,
-} from "@/sse/services/auth";
 import {
  parseEmbeddingModel,
  getAllEmbeddingModels,
-  getEmbeddingProvider,
-  buildDynamicEmbeddingProvider,
-  type EmbeddingProviderNodeRow,
-  type EmbeddingProvider,
 } from "@omniroute/open-sse/config/embeddingRegistry.ts";
-import { errorResponse, unavailableResponse } from "@omniroute/open-sse/utils/error.ts";
+import { errorResponse } from "@omniroute/open-sse/utils/error.ts";
 import { HTTP_STATUS } from "@omniroute/open-sse/config/constants.ts";
 import * as log from "@/sse/utils/logger";
-import { toJsonErrorPayload } from "@/shared/utils/upstreamError";
 import { enforceApiKeyPolicy } from "@/shared/utils/apiKeyPolicy";
 import { v1EmbeddingsSchema } from "@/shared/validation/schemas";
 import { isValidationFailure, validateBody } from "@/shared/validation/helpers";

-import { getAllCustomModels, getProviderNodes, getApiKeyMetadata } from "@/lib/localDb";
+import { getAllCustomModels, getApiKeyMetadata } from "@/lib/localDb";
+import { createEmbeddingResponse, type EmbeddingHandlerOptions } from "@/lib/embeddings/service";
+import { extractApiKey, isValidApiKey } from "@/sse/services/auth";

 function toProviderScopedModelId(providerId: string, modelId: string): string {
  return modelId.startsWith(`${providerId}/`) ? modelId : `${providerId}/${modelId}`;
 }

-/**
- * Handle CORS preflight
- */
 export async function OPTIONS() {
  return new Response(null, {
    headers: {
@@ -39,9 +26,6 @@ export async function OPTIONS() {
  });
 }

-/**
- * GET /v1/embeddings — list available embedding models
- */
 export async function GET() {
  const builtInModels = getAllEmbeddingModels();
  const timestamp = Math.floor(Date.now() / 1000);
@@ -55,7 +39,6 @@ export async function GET() {
    dimensions: m.dimensions,
  }));

-  // Include custom models tagged for embeddings
  try {
    const customModelsMap = (await getAllCustomModels()) as Record<string, any>;
    for (const [providerId, models] of Object.entries(customModelsMap)) {
@@ -82,163 +65,13 @@ export async function GET() {
  });
 }

-/**
- * POST /v1/embeddings — create embeddings
- */
 type ValidatedEmbeddingBody = Record<string, unknown> & { model: string };

-interface EmbeddingHandlerOptions {
-  clientRawRequest?: {
-    endpoint: string;
-    body: Record<string, unknown>;
-    headers: Record<string, string>;
-  };
-  apiKeyId?: string | null;
-  apiKeyName?: string | null;
-  connectionId?: string | null;
-}
-
 export async function handleValidatedEmbeddingRequestBody(
  body: ValidatedEmbeddingBody,
  options: EmbeddingHandlerOptions = {}
 ) {
-  // Load local provider_nodes for embedding routing (only localhost — prevents auth bypass/SSRF)
-  let dynamicProviders: ReturnType<typeof buildDynamicEmbeddingProvider>[] = [];
-  try {
-    const nodes = (await getProviderNodes()) as unknown as EmbeddingProviderNodeRow[];
-    dynamicProviders = (Array.isArray(nodes) ? nodes : [])
-      .filter((n) => {
-        // provider_nodes apiType is "chat", "responses" or "embeddings" — local OpenAI-compatible
-        // backends expose /embeddings under the same base URL as chat, so we build the URL as baseUrl + /embeddings.
-        const validTypes = ["chat", "responses", "embeddings"];
-        if (!validTypes.includes(n.apiType || "")) return false;
-        try {
-          const hostname = new URL(n.baseUrl).hostname;
-          // Strictly matching 172.16.0.0/12 (Docker/local) and explicitly blocking ::1 per SSRF hardening
-          return (
-            hostname === "localhost" ||
-            hostname === "127.0.0.1" ||
-            /^172\.(1[6-9]|2[0-9]|3[0-1])\.\d{1,3}\.\d{1,3}$/.test(hostname)
-          );
-        } catch {
-          return false;
-        }
-      })
-      .map((n) => {
-        try {
-          return buildDynamicEmbeddingProvider(n);
-        } catch (err) {
-          log.error("EMBED", `Skipping invalid provider_node ${n.prefix}: ${err}`);
-          return null;
-        }
-      })
-      .filter((p): p is NonNullable<typeof p> => p !== null);
-  } catch (err) {
-    log.error("EMBED", `Failed to load provider_nodes for embeddings: ${err}`);
-  }
-
-  // Parse model to get provider
-  const { provider, model: resolvedModel } = parseEmbeddingModel(body.model, dynamicProviders);
-  if (!provider) {
-    return errorResponse(
-      HTTP_STATUS.BAD_REQUEST,
-      `Invalid embedding model: ${body.model}. Use format: provider/model`
-    );
-  }
-
-  // Resolve provider config — dynamic first (local override), then hardcoded
-  let providerConfig: EmbeddingProvider | null =
-    dynamicProviders.find((dp) => dp.id === provider) || getEmbeddingProvider(provider) || null;
-  let credentialsProviderId = provider;
-
-  // #496: Fallback — resolve from ALL provider_nodes (not just localhost)
-  // This enables custom embedding models (e.g. google/gemini-embedding-001) whose
-  // providers have remote baseUrls. Safe because getProviderCredentials() authenticates.
-  if (!providerConfig) {
-    try {
-      const allNodes = (await getProviderNodes()) as unknown as EmbeddingProviderNodeRow[];
-      const matchingNode = (Array.isArray(allNodes) ? allNodes : []).find(
-        (n) =>
-          n.prefix === provider &&
-          (n.apiType === "chat" || n.apiType === "responses" || n.apiType === "embeddings") &&
-          n.baseUrl
-      );
-      if (matchingNode) {
-        const baseUrl = String(matchingNode.baseUrl).replace(/\/+$/, "");
-        providerConfig = {
-          id: matchingNode.prefix,
-          baseUrl: `${baseUrl}/embeddings`,
-          authType: "apikey",
-          authHeader: "bearer",
-          models: [],
-        };
-        credentialsProviderId = matchingNode.id || provider;
-        log.info(
-          "EMBED",
-          `Resolved custom embedding provider: ${provider} → ${providerConfig.baseUrl}`
-        );
-      }
-    } catch (err) {
-      log.error("EMBED", `Failed to resolve custom embedding provider ${provider}: ${err}`);
-    }
-  }
-
-  if (!providerConfig) {
-    return errorResponse(
-      HTTP_STATUS.BAD_REQUEST,
-      `Unknown embedding provider: ${provider}. No matching hardcoded or local provider found.`
-    );
-  }
-
-  // Get credentials — skip for local providers (authType: "none")
-  let credentials = null;
-  if (providerConfig && providerConfig.authType !== "none") {
-    credentials = await getProviderCredentials(credentialsProviderId);
-    if (!credentials) {
-      return errorResponse(
-        HTTP_STATUS.BAD_REQUEST,
-        `No credentials for embedding provider: ${provider}`
-      );
-    }
-    if (credentials.allRateLimited) {
-      return unavailableResponse(
-        HTTP_STATUS.RATE_LIMITED,
-        `[${provider}] All accounts rate limited`,
-        credentials.retryAfter,
-        credentials.retryAfterHuman
-      );
-    }
-  }
-
-  const result = await handleEmbedding({
-    body,
-    credentials,
-    log,
-    resolvedProvider: providerConfig,
-    resolvedModel,
-    clientRawRequest: options.clientRawRequest || null,
-    apiKeyId: options.apiKeyId || null,
-    apiKeyName: options.apiKeyName || null,
-    connectionId: options.connectionId || null,
-  });
-
-  const responseHeaders = new Headers(result.headers);
-
-  if (result.success) {
-    if (credentials) await clearRecoveredProviderState(credentials);
-    responseHeaders.set("Content-Type", "application/json");
-    return new Response(JSON.stringify(result.data), {
-      status: result.status,
-      headers: responseHeaders,
-    });
-  }
-
-  responseHeaders.set("Content-Type", "application/json");
-  const errorPayload = toJsonErrorPayload(result.error, "Embedding provider error");
-  return new Response(JSON.stringify(errorPayload), {
-    status: result.status,
-    headers: responseHeaders,
-  });
+  return createEmbeddingResponse(body, options);
 }

 export async function POST(request) {
--- a/src/i18n/messages/ar.json
+++ b/src/i18n/messages/ar.json
@@ -3315,7 +3315,44 @@
    "requestBodyLimitSaveFailed": "Failed to save request body limit",
    "requestBodyLimitSaving": "Saving...",
    "requestBodyLimitSave": "Save",
-    "requestBodyLimitCurrent": "Current: {value}"
+    "requestBodyLimitCurrent": "Current: {value}",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/bg.json
+++ b/src/i18n/messages/bg.json
@@ -3315,7 +3315,44 @@
    "compressionModeRtk": "RTK",
    "compressionModeRtkDesc": "Command-aware tool output filtering",
    "compressionModeStacked": "Stacked",
-    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression"
+    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/bn.json
+++ b/src/i18n/messages/bn.json
@@ -3472,7 +3472,44 @@
    "requestBodyLimitSaveFailed": "Failed to save request body limit",
    "requestBodyLimitSaving": "Saving...",
    "requestBodyLimitSave": "Save",
-    "requestBodyLimitCurrent": "Current: {value}"
+    "requestBodyLimitCurrent": "Current: {value}",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/cs.json
+++ b/src/i18n/messages/cs.json
@@ -3315,7 +3315,44 @@
    "compressionModeRtk": "RTK",
    "compressionModeRtkDesc": "Command-aware tool output filtering",
    "compressionModeStacked": "Stacked",
-    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression"
+    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/da.json
+++ b/src/i18n/messages/da.json
@@ -3315,7 +3315,44 @@
    "requestBodyLimitSaveFailed": "Failed to save request body limit",
    "requestBodyLimitSaving": "Saving...",
    "requestBodyLimitSave": "Save",
-    "requestBodyLimitCurrent": "Current: {value}"
+    "requestBodyLimitCurrent": "Current: {value}",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/de.json
+++ b/src/i18n/messages/de.json
@@ -3331,7 +3331,44 @@
    "requestBodyLimitSaveFailed": "Failed to save request body limit",
    "requestBodyLimitSaving": "Saving...",
    "requestBodyLimitSave": "Save",
-    "requestBodyLimitCurrent": "Current: {value}"
+    "requestBodyLimitCurrent": "Current: {value}",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/en.json
+++ b/src/i18n/messages/en.json
@@ -3758,7 +3758,44 @@
    "compressionModeRtk": "RTK",
    "compressionModeRtkDesc": "Command-aware tool output filtering",
    "compressionModeStacked": "Stacked",
-    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression"
+    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/es.json
+++ b/src/i18n/messages/es.json
@@ -3315,7 +3315,44 @@
    "requestBodyLimitSaveFailed": "Failed to save request body limit",
    "requestBodyLimitSaving": "Saving...",
    "requestBodyLimitSave": "Save",
-    "requestBodyLimitCurrent": "Current: {value}"
+    "requestBodyLimitCurrent": "Current: {value}",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/fa.json
+++ b/src/i18n/messages/fa.json
@@ -3472,7 +3472,44 @@
    "requestBodyLimitSaveFailed": "Failed to save request body limit",
    "requestBodyLimitSaving": "Saving...",
    "requestBodyLimitSave": "Save",
-    "requestBodyLimitCurrent": "Current: {value}"
+    "requestBodyLimitCurrent": "Current: {value}",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/fi.json
+++ b/src/i18n/messages/fi.json
@@ -3315,7 +3315,44 @@
    "compressionModeRtk": "RTK",
    "compressionModeRtkDesc": "Command-aware tool output filtering",
    "compressionModeStacked": "Stacked",
-    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression"
+    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/fr.json
+++ b/src/i18n/messages/fr.json
@@ -3315,7 +3315,44 @@
    "requestBodyLimitSaveFailed": "Failed to save request body limit",
    "requestBodyLimitSaving": "Saving...",
    "requestBodyLimitSave": "Save",
-    "requestBodyLimitCurrent": "Current: {value}"
+    "requestBodyLimitCurrent": "Current: {value}",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/gu.json
+++ b/src/i18n/messages/gu.json
@@ -3472,7 +3472,44 @@
    "requestBodyLimitSaveFailed": "Failed to save request body limit",
    "requestBodyLimitSaving": "Saving...",
    "requestBodyLimitSave": "Save",
-    "requestBodyLimitCurrent": "Current: {value}"
+    "requestBodyLimitCurrent": "Current: {value}",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/he.json
+++ b/src/i18n/messages/he.json
@@ -3315,7 +3315,44 @@
    "compressionModeRtk": "RTK",
    "compressionModeRtkDesc": "Command-aware tool output filtering",
    "compressionModeStacked": "Stacked",
-    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression"
+    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/hi.json
+++ b/src/i18n/messages/hi.json
@@ -3315,7 +3315,44 @@
    "compressionModeRtk": "RTK",
    "compressionModeRtkDesc": "Command-aware tool output filtering",
    "compressionModeStacked": "Stacked",
-    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression"
+    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/hu.json
+++ b/src/i18n/messages/hu.json
@@ -3315,7 +3315,44 @@
    "compressionModeRtk": "RTK",
    "compressionModeRtkDesc": "Command-aware tool output filtering",
    "compressionModeStacked": "Stacked",
-    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression"
+    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/id.json
+++ b/src/i18n/messages/id.json
@@ -3319,7 +3319,44 @@
    "compressionModeRtk": "RTK",
    "compressionModeRtkDesc": "Command-aware tool output filtering",
    "compressionModeStacked": "Stacked",
-    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression"
+    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/in.json
+++ b/src/i18n/messages/in.json
@@ -3472,7 +3472,44 @@
    "requestBodyLimitSaveFailed": "Failed to save request body limit",
    "requestBodyLimitSaving": "Saving...",
    "requestBodyLimitSave": "Save",
-    "requestBodyLimitCurrent": "Current: {value}"
+    "requestBodyLimitCurrent": "Current: {value}",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/it.json
+++ b/src/i18n/messages/it.json
@@ -3315,7 +3315,44 @@
    "requestBodyLimitSaveFailed": "Failed to save request body limit",
    "requestBodyLimitSaving": "Saving...",
    "requestBodyLimitSave": "Save",
-    "requestBodyLimitCurrent": "Current: {value}"
+    "requestBodyLimitCurrent": "Current: {value}",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/ja.json
+++ b/src/i18n/messages/ja.json
@@ -3315,7 +3315,44 @@
    "compressionModeRtk": "RTK",
    "compressionModeRtkDesc": "Command-aware tool output filtering",
    "compressionModeStacked": "Stacked",
-    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression"
+    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/ko.json
+++ b/src/i18n/messages/ko.json
@@ -3317,7 +3317,44 @@
    "compressionModeRtk": "RTK",
    "compressionModeRtkDesc": "Command-aware tool output filtering",
    "compressionModeStacked": "Stacked",
-    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression"
+    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/mr.json
+++ b/src/i18n/messages/mr.json
@@ -3472,7 +3472,44 @@
    "requestBodyLimitSaveFailed": "Failed to save request body limit",
    "requestBodyLimitSaving": "Saving...",
    "requestBodyLimitSave": "Save",
-    "requestBodyLimitCurrent": "Current: {value}"
+    "requestBodyLimitCurrent": "Current: {value}",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/ms.json
+++ b/src/i18n/messages/ms.json
@@ -3315,7 +3315,44 @@
    "requestBodyLimitSaveFailed": "Failed to save request body limit",
    "requestBodyLimitSaving": "Saving...",
    "requestBodyLimitSave": "Save",
-    "requestBodyLimitCurrent": "Current: {value}"
+    "requestBodyLimitCurrent": "Current: {value}",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/nl.json
+++ b/src/i18n/messages/nl.json
@@ -3315,7 +3315,44 @@
    "compressionModeRtk": "RTK",
    "compressionModeRtkDesc": "Command-aware tool output filtering",
    "compressionModeStacked": "Stacked",
-    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression"
+    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/no.json
+++ b/src/i18n/messages/no.json
@@ -3315,7 +3315,44 @@
    "requestBodyLimitSaveFailed": "Failed to save request body limit",
    "requestBodyLimitSaving": "Saving...",
    "requestBodyLimitSave": "Save",
-    "requestBodyLimitCurrent": "Current: {value}"
+    "requestBodyLimitCurrent": "Current: {value}",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/phi.json
+++ b/src/i18n/messages/phi.json
@@ -3315,7 +3315,44 @@
    "compressionModeRtk": "RTK",
    "compressionModeRtkDesc": "Command-aware tool output filtering",
    "compressionModeStacked": "Stacked",
-    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression"
+    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/pl.json
+++ b/src/i18n/messages/pl.json
@@ -3315,7 +3315,44 @@
    "requestBodyLimitSaveFailed": "Failed to save request body limit",
    "requestBodyLimitSaving": "Saving...",
    "requestBodyLimitSave": "Save",
-    "requestBodyLimitCurrent": "Current: {value}"
+    "requestBodyLimitCurrent": "Current: {value}",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/pt-BR.json
+++ b/src/i18n/messages/pt-BR.json
@@ -3432,7 +3432,44 @@
    "compressionModeRtk": "RTK",
    "compressionModeRtkDesc": "Command-aware tool output filtering",
    "compressionModeStacked": "Stacked",
-    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression"
+    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression",
+    "qdrantTitle": "Qdrant (Memoria vetorial)",
+    "qdrantDesc": "Opcional. Indexa memorias semanticas em um banco vetorial externo para busca mais rapida.",
+    "qdrantStatusActive": "Ativo",
+    "qdrantStatusError": "Com erro",
+    "qdrantStatusDisabled": "Desativado",
+    "qdrantEnable": "Ativar Qdrant",
+    "qdrantEnableDesc": "Quando ativo, a estrategia semantic/hybrid pode usar Qdrant para recuperar memorias.",
+    "qdrantTesting": "Testando...",
+    "qdrantTestConnection": "Testar conexao",
+    "qdrantSaved": "Configuracao salva",
+    "qdrantSaveError": "Falha ao salvar configuracao",
+    "qdrantHostHint": "Sem a porta. Ex: 127.0.0.1 ou http://qdrant",
+    "qdrantPort": "Porta",
+    "qdrantPortHint": "Padrao do Qdrant: 6333",
+    "qdrantCollectionHint": "Onde os pontos de memoria serao gravados.",
+    "qdrantEmbeddingModel": "Modelo de embedding",
+    "qdrantHelpTitle": "Ajuda rapida de configuracao",
+    "qdrantHelpQuickTitle": "Configuracao rapida (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: IP/URL do Qdrant, Porta: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. Se usar nvidia/llama-nemotron-embed-vl-1b-v2:free, use dimensao 2048 na collection.",
+    "qdrantHelpStep3": "3. Modelo no campo: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Salvar, testar conexao e depois testar busca.",
+    "qdrantEmbeddingQuickSelect": "Selecao rapida de modelos descobertos...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Formato: provider/model. Precisa ter credencial desse provider configurada.",
+    "qdrantApiKeyPlaceholderKeep": "(deixe vazio para manter)",
+    "qdrantApiKeyPlaceholderOptional": "(deixe vazio se nao usar)",
+    "qdrantSaveHint": "Dica: edite host/porta/collection/modelo e clique em Salvar. A chave e opcional.",
+    "qdrantSearchTestTitle": "Teste de busca",
+    "qdrantSearchTestDesc": "Gera embedding e faz search no Qdrant.",
+    "qdrantSearchPlaceholder": "Ex: preferencias do usuario, historico, etc",
+    "qdrantNoResults": "Sem resultados (ou Qdrant desconfigurado).",
+    "qdrantCleanupTitle": "Retencao e limpeza",
+    "qdrantCleanupDesc": "Remove pontos expirados e antigos, baseado em",
+    "searching": "Buscando...",
+    "cleaning": "Limpando...",
+    "cleanNow": "Limpar agora"
  },
  "contextRtk": {
    "title": "Motor RTK",
--- a/src/i18n/messages/pt.json
+++ b/src/i18n/messages/pt.json
@@ -3402,7 +3402,44 @@
    "requestBodyLimitSaveFailed": "Failed to save request body limit",
    "requestBodyLimitSaving": "Saving...",
    "requestBodyLimitSave": "Save",
-    "requestBodyLimitCurrent": "Current: {value}"
+    "requestBodyLimitCurrent": "Current: {value}",
+    "qdrantTitle": "Qdrant (Memoria vetorial)",
+    "qdrantDesc": "Opcional. Indexa memorias semanticas em um banco vetorial externo para busca mais rapida.",
+    "qdrantStatusActive": "Ativo",
+    "qdrantStatusError": "Com erro",
+    "qdrantStatusDisabled": "Desativado",
+    "qdrantEnable": "Ativar Qdrant",
+    "qdrantEnableDesc": "Quando ativo, a estrategia semantic/hybrid pode usar Qdrant para recuperar memorias.",
+    "qdrantTesting": "Testando...",
+    "qdrantTestConnection": "Testar conexao",
+    "qdrantSaved": "Configuracao salva",
+    "qdrantSaveError": "Falha ao salvar configuracao",
+    "qdrantHostHint": "Sem a porta. Ex: 127.0.0.1 ou http://qdrant",
+    "qdrantPort": "Porta",
+    "qdrantPortHint": "Padrao do Qdrant: 6333",
+    "qdrantCollectionHint": "Onde os pontos de memoria serao gravados.",
+    "qdrantEmbeddingModel": "Modelo de embedding",
+    "qdrantHelpTitle": "Ajuda rapida de configuracao",
+    "qdrantHelpQuickTitle": "Configuracao rapida (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: IP/URL do Qdrant, Porta: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. Se usar nvidia/llama-nemotron-embed-vl-1b-v2:free, use dimensao 2048 na collection.",
+    "qdrantHelpStep3": "3. Modelo no campo: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Salvar, testar conexao e depois testar busca.",
+    "qdrantEmbeddingQuickSelect": "Selecao rapida de modelos descobertos...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Formato: provider/model. Precisa ter credencial desse provider configurada.",
+    "qdrantApiKeyPlaceholderKeep": "(deixe vazio para manter)",
+    "qdrantApiKeyPlaceholderOptional": "(deixe vazio se nao usar)",
+    "qdrantSaveHint": "Dica: edite host/porta/collection/modelo e clique em Salvar. A chave e opcional.",
+    "qdrantSearchTestTitle": "Teste de busca",
+    "qdrantSearchTestDesc": "Gera embedding e faz search no Qdrant.",
+    "qdrantSearchPlaceholder": "Ex: preferencias do usuario, historico, etc",
+    "qdrantNoResults": "Sem resultados (ou Qdrant desconfigurado).",
+    "qdrantCleanupTitle": "Retencao e limpeza",
+    "qdrantCleanupDesc": "Remove pontos expirados e antigos, baseado em",
+    "searching": "Buscando...",
+    "cleaning": "Limpando...",
+    "cleanNow": "Limpar agora"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/ro.json
+++ b/src/i18n/messages/ro.json
@@ -3315,7 +3315,44 @@
    "compressionModeRtk": "RTK",
    "compressionModeRtkDesc": "Command-aware tool output filtering",
    "compressionModeStacked": "Stacked",
-    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression"
+    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/ru.json
+++ b/src/i18n/messages/ru.json
@@ -3339,7 +3339,44 @@
    "compressionModeRtk": "RTK",
    "compressionModeRtkDesc": "Command-aware tool output filtering",
    "compressionModeStacked": "Stacked",
-    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression"
+    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/sk.json
+++ b/src/i18n/messages/sk.json
@@ -3315,7 +3315,44 @@
    "compressionModeRtk": "RTK",
    "compressionModeRtkDesc": "Command-aware tool output filtering",
    "compressionModeStacked": "Stacked",
-    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression"
+    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/sv.json
+++ b/src/i18n/messages/sv.json
@@ -3315,7 +3315,44 @@
    "compressionModeRtk": "RTK",
    "compressionModeRtkDesc": "Command-aware tool output filtering",
    "compressionModeStacked": "Stacked",
-    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression"
+    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/sw.json
+++ b/src/i18n/messages/sw.json
@@ -3472,7 +3472,44 @@
    "requestBodyLimitSaveFailed": "Failed to save request body limit",
    "requestBodyLimitSaving": "Saving...",
    "requestBodyLimitSave": "Save",
-    "requestBodyLimitCurrent": "Current: {value}"
+    "requestBodyLimitCurrent": "Current: {value}",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/ta.json
+++ b/src/i18n/messages/ta.json
@@ -3472,7 +3472,44 @@
    "requestBodyLimitSaveFailed": "Failed to save request body limit",
    "requestBodyLimitSaving": "Saving...",
    "requestBodyLimitSave": "Save",
-    "requestBodyLimitCurrent": "Current: {value}"
+    "requestBodyLimitCurrent": "Current: {value}",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/te.json
+++ b/src/i18n/messages/te.json
@@ -3472,7 +3472,44 @@
    "requestBodyLimitSaveFailed": "Failed to save request body limit",
    "requestBodyLimitSaving": "Saving...",
    "requestBodyLimitSave": "Save",
-    "requestBodyLimitCurrent": "Current: {value}"
+    "requestBodyLimitCurrent": "Current: {value}",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/th.json
+++ b/src/i18n/messages/th.json
@@ -3315,7 +3315,44 @@
    "requestBodyLimitSaveFailed": "Failed to save request body limit",
    "requestBodyLimitSaving": "Saving...",
    "requestBodyLimitSave": "Save",
-    "requestBodyLimitCurrent": "Current: {value}"
+    "requestBodyLimitCurrent": "Current: {value}",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/tr.json
+++ b/src/i18n/messages/tr.json
@@ -3315,7 +3315,44 @@
    "compressionModeRtk": "RTK",
    "compressionModeRtkDesc": "Command-aware tool output filtering",
    "compressionModeStacked": "Stacked",
-    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression"
+    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/uk-UA.json
+++ b/src/i18n/messages/uk-UA.json
@@ -3315,7 +3315,44 @@
    "compressionModeRtk": "RTK",
    "compressionModeRtkDesc": "Command-aware tool output filtering",
    "compressionModeStacked": "Stacked",
-    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression"
+    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/ur.json
+++ b/src/i18n/messages/ur.json
@@ -3472,7 +3472,44 @@
    "requestBodyLimitSaveFailed": "Failed to save request body limit",
    "requestBodyLimitSaving": "Saving...",
    "requestBodyLimitSave": "Save",
-    "requestBodyLimitCurrent": "Current: {value}"
+    "requestBodyLimitCurrent": "Current: {value}",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/vi.json
+++ b/src/i18n/messages/vi.json
@@ -3315,7 +3315,44 @@
    "requestBodyLimitSaveFailed": "Failed to save request body limit",
    "requestBodyLimitSaving": "Saving...",
    "requestBodyLimitSave": "Save",
-    "requestBodyLimitCurrent": "Current: {value}"
+    "requestBodyLimitCurrent": "Current: {value}",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/i18n/messages/zh-CN.json
+++ b/src/i18n/messages/zh-CN.json
@@ -3424,7 +3424,44 @@
    "compressionModeRtk": "RTK",
    "compressionModeRtkDesc": "Command-aware tool output filtering",
    "compressionModeStacked": "Stacked",
-    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression"
+    "compressionModeStackedDesc": "RTK tool-output filtering followed by Caveman message compression",
+    "qdrantTitle": "Qdrant (Vector memory)",
+    "qdrantDesc": "Optional. Indexes semantic memories in an external vector database for faster retrieval.",
+    "qdrantStatusActive": "Active",
+    "qdrantStatusError": "Error",
+    "qdrantStatusDisabled": "Disabled",
+    "qdrantEnable": "Enable Qdrant",
+    "qdrantEnableDesc": "When enabled, semantic/hybrid strategy can use Qdrant to retrieve memories.",
+    "qdrantTesting": "Testing...",
+    "qdrantTestConnection": "Test connection",
+    "qdrantSaved": "Configuration saved",
+    "qdrantSaveError": "Failed to save configuration",
+    "qdrantHostHint": "Without port. Example: 127.0.0.1 or http://qdrant",
+    "qdrantPort": "Port",
+    "qdrantPortHint": "Qdrant default: 6333",
+    "qdrantCollectionHint": "Where memory points will be stored.",
+    "qdrantEmbeddingModel": "Embedding model",
+    "qdrantHelpTitle": "Quick setup help",
+    "qdrantHelpQuickTitle": "Quick setup (Qdrant + OpenRouter)",
+    "qdrantHelpStep1": "1. Host: Qdrant IP/URL, Port: 6333, Collection: omniroute_memory.",
+    "qdrantHelpStep2": "2. If using nvidia/llama-nemotron-embed-vl-1b-v2:free, use collection dimension 2048.",
+    "qdrantHelpStep3": "3. Model field: openrouter/nvidia/llama-nemotron-embed-vl-1b-v2:free.",
+    "qdrantHelpStep4": "4. Save, test connection, then test search.",
+    "qdrantEmbeddingQuickSelect": "Quick select from discovered models...",
+    "qdrantEmbeddingInputPlaceholder": "openai/text-embedding-3-small",
+    "qdrantEmbeddingHint": "Format: provider/model. The provider credential must be configured.",
+    "qdrantApiKeyPlaceholderKeep": "(leave empty to keep current key)",
+    "qdrantApiKeyPlaceholderOptional": "(leave empty if not used)",
+    "qdrantSaveHint": "Tip: edit host/port/collection/model and click Save. API key is optional.",
+    "qdrantSearchTestTitle": "Search test",
+    "qdrantSearchTestDesc": "Generates embedding and searches in Qdrant.",
+    "qdrantSearchPlaceholder": "Example: user preferences, history, etc",
+    "qdrantNoResults": "No results (or Qdrant is not configured).",
+    "qdrantCleanupTitle": "Retention and cleanup",
+    "qdrantCleanupDesc": "Removes expired and old points based on",
+    "searching": "Searching...",
+    "cleaning": "Cleaning...",
+    "cleanNow": "Clean now"
  },
  "contextRtk": {
    "title": "RTK Engine",
--- a/src/lib/embeddings/service.ts
+++ b/src/lib/embeddings/service.ts
@@ -0,0 +1,160 @@
+import { handleEmbedding } from "@omniroute/open-sse/handlers/embeddings.ts";
+import {
+  parseEmbeddingModel,
+  getEmbeddingProvider,
+  buildDynamicEmbeddingProvider,
+  type EmbeddingProviderNodeRow,
+  type EmbeddingProvider,
+} from "@omniroute/open-sse/config/embeddingRegistry.ts";
+import { errorResponse, unavailableResponse } from "@omniroute/open-sse/utils/error.ts";
+import { HTTP_STATUS } from "@omniroute/open-sse/config/constants.ts";
+import * as log from "@/sse/utils/logger";
+import { toJsonErrorPayload } from "@/shared/utils/upstreamError";
+import { getProviderCredentials, clearRecoveredProviderState } from "@/sse/services/auth";
+import { getProviderNodes } from "@/lib/localDb";
+
+type ValidatedEmbeddingBody = Record<string, unknown> & { model: string };
+
+interface EmbeddingHandlerOptions {
+  clientRawRequest?: {
+    endpoint: string;
+    body: Record<string, unknown>;
+    headers: Record<string, string>;
+  };
+  apiKeyId?: string | null;
+  apiKeyName?: string | null;
+  connectionId?: string | null;
+}
+
+export async function createEmbeddingResponse(
+  body: ValidatedEmbeddingBody,
+  options: EmbeddingHandlerOptions = {}
+): Promise<Response> {
+  let dynamicProviders: ReturnType<typeof buildDynamicEmbeddingProvider>[] = [];
+  try {
+    const nodes = (await getProviderNodes()) as unknown as EmbeddingProviderNodeRow[];
+    dynamicProviders = (Array.isArray(nodes) ? nodes : [])
+      .filter((n) => {
+        const validTypes = ["chat", "responses", "embeddings"];
+        if (!validTypes.includes(n.apiType || "")) return false;
+        try {
+          const hostname = new URL(n.baseUrl).hostname;
+          return (
+            hostname === "localhost" ||
+            hostname === "127.0.0.1" ||
+            /^172\.(1[6-9]|2[0-9]|3[0-1])\.\d{1,3}\.\d{1,3}$/.test(hostname)
+          );
+        } catch {
+          return false;
+        }
+      })
+      .map((n) => {
+        try {
+          return buildDynamicEmbeddingProvider(n);
+        } catch (err) {
+          log.error("EMBED", `Skipping invalid provider_node ${n.prefix}: ${err}`);
+          return null;
+        }
+      })
+      .filter((p): p is NonNullable<typeof p> => p !== null);
+  } catch (err) {
+    log.error("EMBED", `Failed to load provider_nodes for embeddings: ${err}`);
+  }
+
+  const { provider, model: resolvedModel } = parseEmbeddingModel(body.model, dynamicProviders);
+  if (!provider) {
+    return errorResponse(
+      HTTP_STATUS.BAD_REQUEST,
+      `Invalid embedding model: ${body.model}. Use format: provider/model`
+    );
+  }
+
+  let providerConfig: EmbeddingProvider | null =
+    dynamicProviders.find((dp) => dp.id === provider) || getEmbeddingProvider(provider) || null;
+  let credentialsProviderId = provider;
+
+  if (!providerConfig) {
+    try {
+      const allNodes = (await getProviderNodes()) as unknown as EmbeddingProviderNodeRow[];
+      const matchingNode = (Array.isArray(allNodes) ? allNodes : []).find(
+        (n) =>
+          n.prefix === provider &&
+          (n.apiType === "chat" || n.apiType === "responses" || n.apiType === "embeddings") &&
+          n.baseUrl
+      );
+      if (matchingNode) {
+        const baseUrl = String(matchingNode.baseUrl).replace(/\/+$/, "");
+        providerConfig = {
+          id: matchingNode.prefix,
+          baseUrl: `${baseUrl}/embeddings`,
+          authType: "apikey",
+          authHeader: "bearer",
+          models: [],
+        };
+        credentialsProviderId = matchingNode.id || provider;
+        log.info(
+          "EMBED",
+          `Resolved custom embedding provider: ${provider} -> ${providerConfig.baseUrl}`
+        );
+      }
+    } catch (err) {
+      log.error("EMBED", `Failed to resolve custom embedding provider ${provider}: ${err}`);
+    }
+  }
+
+  if (!providerConfig) {
+    return errorResponse(
+      HTTP_STATUS.BAD_REQUEST,
+      `Unknown embedding provider: ${provider}. No matching hardcoded or local provider found.`
+    );
+  }
+
+  let credentials = null;
+  if (providerConfig.authType !== "none") {
+    credentials = await getProviderCredentials(credentialsProviderId);
+    if (!credentials) {
+      return errorResponse(
+        HTTP_STATUS.BAD_REQUEST,
+        `No credentials for embedding provider: ${provider}`
+      );
+    }
+    if (credentials.allRateLimited) {
+      return unavailableResponse(
+        HTTP_STATUS.RATE_LIMITED,
+        `[${provider}] All accounts rate limited`,
+        credentials.retryAfter,
+        credentials.retryAfterHuman
+      );
+    }
+  }
+
+  const result = await handleEmbedding({
+    body,
+    credentials,
+    log,
+    resolvedProvider: providerConfig,
+    resolvedModel,
+    clientRawRequest: options.clientRawRequest || null,
+    apiKeyId: options.apiKeyId || null,
+    apiKeyName: options.apiKeyName || null,
+    connectionId: options.connectionId || null,
+  });
+
+  const responseHeaders = new Headers(result.headers);
+
+  if (result.success) {
+    if (credentials) await clearRecoveredProviderState(credentials);
+    responseHeaders.set("Content-Type", "application/json");
+    return new Response(JSON.stringify(result.data), {
+      status: result.status,
+      headers: responseHeaders,
+    });
+  }
+
+  responseHeaders.set("Content-Type", "application/json");
+  const errorPayload = toJsonErrorPayload(result.error, "Embedding provider error");
+  return new Response(JSON.stringify(errorPayload), {
+    status: result.status,
+    headers: responseHeaders,
+  });
+}
--- a/src/lib/memory/qdrant.ts
+++ b/src/lib/memory/qdrant.ts
@@ -0,0 +1,409 @@
+import { getSettings } from "@/lib/db/settings";
+import { createEmbeddingResponse } from "@/lib/embeddings/service";
+
+type JsonRecord = Record<string, unknown>;
+
+export type QdrantConfig = {
+  enabled: boolean;
+  host: string;
+  port: number;
+  apiKey: string | null;
+  collection: string;
+  embeddingModel: string;
+};
+
+export function normalizeQdrantConfig(settings: Record<string, unknown>): QdrantConfig {
+  const host = typeof settings.qdrantHost === "string" ? settings.qdrantHost.trim() : "";
+  const portRaw = settings.qdrantPort;
+  const port =
+    typeof portRaw === "number" && Number.isFinite(portRaw)
+      ? Math.round(portRaw)
+      : typeof portRaw === "string"
+        ? Math.round(Number(portRaw) || 6333)
+        : 6333;
+  const apiKey =
+    typeof settings.qdrantApiKey === "string" && settings.qdrantApiKey.trim().length > 0
+      ? settings.qdrantApiKey.trim()
+      : null;
+  const collection =
+    typeof settings.qdrantCollection === "string" && settings.qdrantCollection.trim().length > 0
+      ? settings.qdrantCollection.trim()
+      : "omniroute_memory";
+  const embeddingModel =
+    typeof settings.qdrantEmbeddingModel === "string" &&
+    settings.qdrantEmbeddingModel.trim().length > 0
+      ? settings.qdrantEmbeddingModel.trim()
+      : "openai/text-embedding-3-small";
+  const enabled = settings.qdrantEnabled === true;
+
+  return { enabled, host, port, apiKey, collection, embeddingModel };
+}
+
+export async function getQdrantConfig(): Promise<QdrantConfig> {
+  const settings = (await getSettings()) as Record<string, unknown>;
+  return normalizeQdrantConfig(settings);
+}
+
+function baseUrl(cfg: QdrantConfig): string {
+  const host = cfg.host.replace(/\/+$/, "");
+  const withProto =
+    host.startsWith("http://") || host.startsWith("https://") ? host : `http://${host}`;
+  try {
+    const url = new URL(withProto);
+    if (!url.port) url.port = String(cfg.port);
+    return url.toString().replace(/\/+$/, "");
+  } catch {
+    return `${withProto}:${cfg.port}`;
+  }
+}
+
+async function qdrantFetch(cfg: QdrantConfig, path: string, init?: RequestInit): Promise<Response> {
+  const headers: Record<string, string> = {
+    "content-type": "application/json",
+    ...(init?.headers as Record<string, string> | undefined),
+  };
+  if (cfg.apiKey) headers["api-key"] = cfg.apiKey;
+
+  return fetch(`${baseUrl(cfg)}${path}`, {
+    ...init,
+    headers,
+  });
+}
+
+export async function checkQdrantHealth(): Promise<{
+  ok: boolean;
+  latencyMs: number;
+  error?: string;
+}> {
+  const cfg = await getQdrantConfig();
+  const start = Date.now();
+  if (!cfg.enabled || !cfg.host) {
+    return { ok: false, latencyMs: 0, error: "not_configured" };
+  }
+
+  try {
+    const res = await qdrantFetch(cfg, "/readyz", { method: "GET" });
+    const latencyMs = Date.now() - start;
+    if (!res.ok) {
+      const text = await res.text().catch(() => "");
+      return { ok: false, latencyMs, error: text.slice(0, 200) || `HTTP ${res.status}` };
+    }
+    return { ok: true, latencyMs };
+  } catch (err) {
+    return {
+      ok: false,
+      latencyMs: Date.now() - start,
+      error: err instanceof Error ? err.message : String(err),
+    };
+  }
+}
+
+async function ensureCollection(cfg: QdrantConfig, vectorSize: number): Promise<void> {
+  const getRes = await qdrantFetch(cfg, `/collections/${encodeURIComponent(cfg.collection)}`, {
+    method: "GET",
+  });
+  if (getRes.ok) return;
+
+  const createRes = await qdrantFetch(cfg, `/collections/${encodeURIComponent(cfg.collection)}`, {
+    method: "PUT",
+    body: JSON.stringify({
+      vectors: { size: vectorSize, distance: "Cosine" },
+    }),
+  });
+  if (!createRes.ok) {
+    const text = await createRes.text().catch(() => "");
+    throw new Error(text.slice(0, 300) || `Failed to create collection (${createRes.status})`);
+  }
+}
+
+async function getCollectionVectorName(cfg: QdrantConfig): Promise<string | null> {
+  const res = await qdrantFetch(cfg, `/collections/${encodeURIComponent(cfg.collection)}`, {
+    method: "GET",
+  });
+  if (!res.ok) return null;
+  const data = (await res.json().catch(() => null)) as any;
+  const vectors = data?.result?.config?.params?.vectors;
+  if (!vectors || typeof vectors !== "object" || Array.isArray(vectors)) {
+    return null;
+  }
+  // Unnamed/single-vector config: { size, distance, ... } (not a named map)
+  if (
+    Object.prototype.hasOwnProperty.call(vectors, "size") &&
+    (typeof vectors.size === "number" || typeof vectors.size === "string")
+  ) {
+    return null;
+  }
+  const names = Object.keys(vectors);
+  if (names.length === 0) return null;
+  return names[0] || null;
+}
+
+async function embedText(cfg: QdrantConfig, text: string): Promise<number[]> {
+  const modelStr = cfg.embeddingModel.trim();
+  if (!modelStr.includes("/")) {
+    throw new Error(`Invalid embedding model '${modelStr}'. Use provider/model format.`);
+  }
+
+  const res = await createEmbeddingResponse({
+    model: modelStr,
+    input: text,
+  });
+  if (!res.ok) {
+    const txt = await res.text().catch(() => "");
+    throw new Error(txt.slice(0, 300) || `Embeddings request failed (${res.status})`);
+  }
+  const data = (await res.json().catch(() => null)) as any;
+  const vec = data?.data?.[0]?.embedding;
+  if (!Array.isArray(vec) || vec.length === 0) {
+    throw new Error("Embedding response missing vector");
+  }
+  return vec as number[];
+}
+
+export async function upsertSemanticMemoryPoint(input: {
+  id: string;
+  apiKeyId: string;
+  sessionId: string;
+  key: string;
+  content: string;
+  metadata: JsonRecord;
+  createdAt: string;
+  expiresAt: string | null;
+}): Promise<{ ok: boolean; latencyMs: number; error?: string }> {
+  const cfg = await getQdrantConfig();
+  if (!cfg.enabled || !cfg.host) return { ok: false, latencyMs: 0, error: "not_configured" };
+
+  const start = Date.now();
+  try {
+    const vector = await embedText(cfg, `${input.key}\n\n${input.content}`);
+    await ensureCollection(cfg, vector.length);
+    const vectorName = await getCollectionVectorName(cfg);
+
+    const createdAtUnix = Math.floor(new Date(input.createdAt).getTime() / 1000);
+    const expiresAtUnix = input.expiresAt
+      ? Math.floor(new Date(input.expiresAt).getTime() / 1000)
+      : null;
+
+    const payload = {
+      kind: "omniroute_memory",
+      memoryId: input.id,
+      apiKeyId: input.apiKeyId || "",
+      sessionId: input.sessionId || "",
+      type: "semantic",
+      key: input.key || "",
+      content: input.content || "",
+      metadata: input.metadata || {},
+      createdAtUnix,
+      expiresAtUnix,
+    };
+
+    const res = await qdrantFetch(
+      cfg,
+      `/collections/${encodeURIComponent(cfg.collection)}/points?wait=true`,
+      {
+        method: "PUT",
+        body: JSON.stringify({
+          points: [
+            {
+              id: input.id,
+              vector: vectorName ? { [vectorName]: vector } : vector,
+              payload,
+            },
+          ],
+        }),
+      }
+    );
+
+    const latencyMs = Date.now() - start;
+    if (!res.ok) {
+      const text = await res.text().catch(() => "");
+      return { ok: false, latencyMs, error: text.slice(0, 300) || `HTTP ${res.status}` };
+    }
+    return { ok: true, latencyMs };
+  } catch (err) {
+    return {
+      ok: false,
+      latencyMs: Date.now() - start,
+      error: err instanceof Error ? err.message : String(err),
+    };
+  }
+}
+
+export async function searchSemanticMemory(
+  query: string,
+  topK = 5,
+  scope?: { apiKeyId?: string; sessionId?: string | null }
+): Promise<{
+  ok: boolean;
+  latencyMs: number;
+  results?: Array<{ id: string; score: number; payload?: JsonRecord }>;
+  error?: string;
+}> {
+  const cfg = await getQdrantConfig();
+  if (!cfg.enabled || !cfg.host) return { ok: false, latencyMs: 0, error: "not_configured" };
+  const start = Date.now();
+  try {
+    const vector = await embedText(cfg, query);
+    await ensureCollection(cfg, vector.length);
+    const vectorName = await getCollectionVectorName(cfg);
+
+    const res = await qdrantFetch(
+      cfg,
+      `/collections/${encodeURIComponent(cfg.collection)}/points/search`,
+      {
+        method: "POST",
+        body: JSON.stringify({
+          vector: vectorName ? { name: vectorName, vector } : vector,
+          limit: Math.max(1, Math.min(20, topK)),
+          filter: {
+            must: [
+              { key: "kind", match: { value: "omniroute_memory" } },
+              ...(scope?.apiKeyId ? [{ key: "apiKeyId", match: { value: scope.apiKeyId } }] : []),
+              ...(scope?.sessionId
+                ? [{ key: "sessionId", match: { value: String(scope.sessionId) } }]
+                : []),
+            ],
+          },
+          with_payload: true,
+        }),
+      }
+    );
+
+    const latencyMs = Date.now() - start;
+    if (!res.ok) {
+      const text = await res.text().catch(() => "");
+      return { ok: false, latencyMs, error: text.slice(0, 300) || `HTTP ${res.status}` };
+    }
+    const data = (await res.json().catch(() => null)) as any;
+    const result = Array.isArray(data?.result) ? data.result : [];
+    return {
+      ok: true,
+      latencyMs,
+      results: result.map((r: any) => ({
+        id: String(r.id),
+        score: typeof r.score === "number" ? r.score : 0,
+        payload: r.payload && typeof r.payload === "object" ? (r.payload as JsonRecord) : undefined,
+      })),
+    };
+  } catch (err) {
+    return {
+      ok: false,
+      latencyMs: Date.now() - start,
+      error: err instanceof Error ? err.message : String(err),
+    };
+  }
+}
+
+export async function deleteSemanticMemoryPoint(
+  id: string
+): Promise<{ ok: boolean; latencyMs: number; error?: string }> {
+  const cfg = await getQdrantConfig();
+  if (!cfg.enabled || !cfg.host) return { ok: false, latencyMs: 0, error: "not_configured" };
+  const start = Date.now();
+  try {
+    const res = await qdrantFetch(
+      cfg,
+      `/collections/${encodeURIComponent(cfg.collection)}/points/delete?wait=true`,
+      {
+        method: "POST",
+        body: JSON.stringify({ points: [id] }),
+      }
+    );
+    const latencyMs = Date.now() - start;
+    if (!res.ok) {
+      const text = await res.text().catch(() => "");
+      return { ok: false, latencyMs, error: text.slice(0, 300) || `HTTP ${res.status}` };
+    }
+    return { ok: true, latencyMs };
+  } catch (err) {
+    return {
+      ok: false,
+      latencyMs: Date.now() - start,
+      error: err instanceof Error ? err.message : String(err),
+    };
+  }
+}
+
+export async function cleanupSemanticMemoryPoints(input: {
+  retentionDays: number;
+}): Promise<{ ok: boolean; deletedCount: number; latencyMs: number; error?: string }> {
+  const cfg = await getQdrantConfig();
+  if (!cfg.enabled || !cfg.host)
+    return { ok: false, deletedCount: 0, latencyMs: 0, error: "not_configured" };
+
+  const retentionDays =
+    typeof input.retentionDays === "number" && Number.isFinite(input.retentionDays)
+      ? Math.max(1, Math.min(3650, Math.round(input.retentionDays)))
+      : 30;
+
+  const start = Date.now();
+  try {
+    const nowUnix = Math.floor(Date.now() / 1000);
+    const cutoffUnix = nowUnix - retentionDays * 24 * 60 * 60;
+
+    const filter: Record<string, unknown> = {
+      must: [{ key: "kind", match: { value: "omniroute_memory" } }],
+      should: [
+        { key: "expiresAtUnix", range: { lt: nowUnix } },
+        { key: "createdAtUnix", range: { lt: cutoffUnix } },
+      ],
+    };
+
+    // Count first (so we can show an actual number in the dashboard)
+    const countRes = await qdrantFetch(
+      cfg,
+      `/collections/${encodeURIComponent(cfg.collection)}/points/count`,
+      {
+        method: "POST",
+        body: JSON.stringify({ filter, exact: true }),
+      }
+    );
+    if (!countRes.ok) {
+      const text = await countRes.text().catch(() => "");
+      return {
+        ok: false,
+        deletedCount: 0,
+        latencyMs: Date.now() - start,
+        error: text.slice(0, 300) || `HTTP ${countRes.status}`,
+      };
+    }
+    const countData = (await countRes.json().catch(() => null)) as any;
+    const toDelete =
+      typeof countData?.result?.count === "number" && Number.isFinite(countData.result.count)
+        ? Math.max(0, Math.round(countData.result.count))
+        : 0;
+
+    if (toDelete === 0) {
+      return { ok: true, deletedCount: 0, latencyMs: Date.now() - start };
+    }
+
+    const delRes = await qdrantFetch(
+      cfg,
+      `/collections/${encodeURIComponent(cfg.collection)}/points/delete?wait=true`,
+      {
+        method: "POST",
+        body: JSON.stringify({
+          filter,
+        }),
+      }
+    );
+    if (!delRes.ok) {
+      const text = await delRes.text().catch(() => "");
+      return {
+        ok: false,
+        deletedCount: 0,
+        latencyMs: Date.now() - start,
+        error: text.slice(0, 300) || `HTTP ${delRes.status}`,
+      };
+    }
+
+    return { ok: true, deletedCount: toDelete, latencyMs: Date.now() - start };
+  } catch (err) {
+    return {
+      ok: false,
+      deletedCount: 0,
+      latencyMs: Date.now() - start,
+      error: err instanceof Error ? err.message : String(err),
+    };
+  }
+}