diff --git a/changelog.d/fixes/9161-codex-responses-chat-targets.md b/changelog.d/fixes/9161-codex-responses-chat-targets.md
new file mode 100644
index 0000000000..2fff43b6e8
--- /dev/null
+++ b/changelog.d/fixes/9161-codex-responses-chat-targets.md
@@ -0,0 +1 @@
+- **fix(translator):** Honor configured Chat targets for Responses-shaped clients while preserving native Responses providers and outbound token fields ([#9161](https://github.com/diegosouzapw/OmniRoute/pull/9161)) — thanks @Zartharas
diff --git a/open-sse/handlers/chatCore/sanitization.ts b/open-sse/handlers/chatCore/sanitization.ts
index 62b43615ed..292d0a4ef6 100644
--- a/open-sse/handlers/chatCore/sanitization.ts
+++ b/open-sse/handlers/chatCore/sanitization.ts
@@ -6,8 +6,8 @@ export function sanitizeChatRequestBody(
   sourceFormat: string,
   targetFormat: string
 ): Record<string, unknown> {
-  const prefersResponsesTokenField =
-    sourceFormat === FORMATS.OPENAI_RESPONSES || targetFormat === FORMATS.OPENAI_RESPONSES;
+  void sourceFormat;
+  const prefersResponsesTokenField = targetFormat === FORMATS.OPENAI_RESPONSES;
 
   if (prefersResponsesTokenField) {
     if (body.max_output_tokens === undefined) {
diff --git a/open-sse/handlers/chatCore/targetFormat.ts b/open-sse/handlers/chatCore/targetFormat.ts
index 9bb2f0da7a..ee56edd226 100644
--- a/open-sse/handlers/chatCore/targetFormat.ts
+++ b/open-sse/handlers/chatCore/targetFormat.ts
@@ -2,56 +2,31 @@
  * chatCore wire target-format resolver (Quality Gate v2 / Fase 9 — chatCore god-file
  * decomposition, #3501).
  *
- * Pure resolution of the provider alias + the upstream target format used to translate the request:
- * apiFormat==="responses" forces OpenAI Responses; otherwise the model's registry target format, then
- * the per-model custom override (#2905), then AgentRouter's matching inbound protocol when the
- * connection has no explicit override, then the provider default. Returns both `alias` (reused by
- * the handler when stripping the `alias/` prefix off the upstream model id) and `targetFormat`.
- * Side-effect-free; sits alongside the other request-setup resolvers
- * (resolveChatCoreRequestSetup / resolveChatCoreRequestFormat).
+ * Pure resolution of the provider alias + the upstream target format used to translate the request.
+ * The inbound API shape does not determine the outbound protocol: the model registry target format,
+ * per-model custom override (#2905), and provider configuration take precedence. This lets a
+ * Responses-shaped client request be translated to a Chat Completions-compatible upstream while
+ * preserving native Responses providers configured with an OpenAI Responses target.
+ * Returns both `alias` (reused by the handler when stripping the `alias/` prefix off the upstream
+ * model id) and `targetFormat`.
  */
 
 import { PROVIDER_ID_TO_ALIAS, getModelTargetFormat } from "../../config/providerModels.ts";
 import { getTargetFormat } from "../../services/provider.ts";
-import { FORMATS } from "../../translator/formats.ts";
 
 export function resolveChatCoreTargetFormat(opts: {
   provider: string;
   resolvedModel: string;
   apiFormat: string | undefined;
-  sourceFormat?: string;
   customModelTargetFormat: string | undefined;
   providerSpecificData: unknown;
 }) {
-  const {
-    provider,
-    resolvedModel,
-    apiFormat,
-    sourceFormat,
-    customModelTargetFormat,
-    providerSpecificData,
-  } = opts;
+  const { provider, resolvedModel, customModelTargetFormat, providerSpecificData } = opts;
   const alias = PROVIDER_ID_TO_ALIAS[provider] || provider;
   const modelTargetFormat = getModelTargetFormat(alias, resolvedModel);
-  const explicitConnectionTargetFormat = (
-    providerSpecificData as { targetFormat?: unknown } | null | undefined
-  )?.targetFormat;
-  const inferredAgentRouterTargetFormat =
-    provider === "agentrouter" &&
-    !(typeof explicitConnectionTargetFormat === "string" && explicitConnectionTargetFormat) &&
-    (sourceFormat === FORMATS.OPENAI_RESPONSES ||
-      sourceFormat === FORMATS.OPENAI ||
-      sourceFormat === FORMATS.CLAUDE)
-      ? sourceFormat
-      : undefined;
   const targetFormat =
-    apiFormat === "responses"
-      ? FORMATS.OPENAI_RESPONSES
-      : modelTargetFormat ||
-        customModelTargetFormat ||
-        inferredAgentRouterTargetFormat ||
-        getTargetFormat(provider, providerSpecificData);
+    modelTargetFormat || customModelTargetFormat || getTargetFormat(provider, providerSpecificData);
   return { alias, targetFormat };
 }
 
-export type ChatCoreTargetFormat = ReturnType<typeof resolveChatCoreTargetFormat>;
+export type ChatCoreTargetFormat = ReturnType<typeof resolveChatCoreTargetFormat>;
\ No newline at end of file
diff --git a/tests/unit/chatcore-extracted-modules-3821.test.ts b/tests/unit/chatcore-extracted-modules-3821.test.ts
index 009412f98e..2ebecefe87 100644
--- a/tests/unit/chatcore-extracted-modules-3821.test.ts
+++ b/tests/unit/chatcore-extracted-modules-3821.test.ts
@@ -18,8 +18,12 @@ import {
 import { FORMATS } from "../../open-sse/translator/formats.ts";
 import { saveIdempotency } from "../../src/lib/idempotencyLayer.ts";
 
-test("sanitizeChatRequestBody: Chat Completions target maps max_output_tokens → max_tokens", () => {
-  const out = sanitizeChatRequestBody({ max_output_tokens: 256 }, FORMATS.OPENAI, FORMATS.OPENAI);
+test("sanitizeChatRequestBody: Responses source targeting Chat maps max_output_tokens → max_tokens", () => {
+  const out = sanitizeChatRequestBody(
+    { max_output_tokens: 256 },
+    FORMATS.OPENAI_RESPONSES,
+    FORMATS.OPENAI
+  );
   assert.equal(out.max_tokens, 256);
   assert.equal(out.max_output_tokens, undefined);
 });
@@ -35,7 +39,11 @@ test("sanitizeChatRequestBody: Responses target maps max_completion_tokens → m
 });
 
 test("sanitizeChatRequestBody: Responses target maps max_tokens → max_output_tokens", () => {
-  const out = sanitizeChatRequestBody({ max_tokens: 128 }, FORMATS.OPENAI_RESPONSES, FORMATS.OPENAI);
+  const out = sanitizeChatRequestBody(
+    { max_tokens: 128 },
+    FORMATS.OPENAI,
+    FORMATS.OPENAI_RESPONSES
+  );
   assert.equal(out.max_output_tokens, 128);
   assert.equal(out.max_tokens, undefined);
 });
diff --git a/tests/unit/chatcore-target-format.test.ts b/tests/unit/chatcore-target-format.test.ts
index 2fb4da01dd..098d349d61 100644
--- a/tests/unit/chatcore-target-format.test.ts
+++ b/tests/unit/chatcore-target-format.test.ts
@@ -1,72 +1,85 @@
 // tests/unit/chatcore-target-format.test.ts
 // Characterization of resolveChatCoreTargetFormat — the wire target-format resolution extracted
-// from handleChatCore (chatCore god-file decomposition, #3501). Resolves the provider alias and the
-// upstream target format: apiFormat==="responses" forces OpenAI Responses; otherwise the model's
-// registry target format, then the custom-model override, then the provider default. Returns both
-// `alias` (reused downstream when stripping the alias/ prefix off the upstream model) and
-// `targetFormat`.
+// from handleChatCore (chatCore god-file decomposition, #3501). The inbound client API shape is
+// independent from the outbound provider protocol: model registry metadata, custom-model overrides,
+// and provider configuration determine the upstream target format.
 import { test } from "node:test";
 import assert from "node:assert/strict";
 import { resolveChatCoreTargetFormat } from "../../open-sse/handlers/chatCore/targetFormat.ts";
-import {
-  PROVIDER_ID_TO_ALIAS,
-  getModelTargetFormat,
-} from "../../open-sse/config/providerModels.ts";
+import { PROVIDER_ID_TO_ALIAS, getModelTargetFormat } from "../../open-sse/config/providerModels.ts";
 import { getTargetFormat } from "../../open-sse/services/provider.ts";
 import { FORMATS } from "../../open-sse/translator/formats.ts";
 
 function expected(
   provider: string,
   resolvedModel: string,
-  apiFormat: string | undefined,
   customModelTargetFormat: string | undefined,
   providerSpecificData: unknown
 ) {
   const alias = PROVIDER_ID_TO_ALIAS[provider] || provider;
   const modelTargetFormat = getModelTargetFormat(alias, resolvedModel);
   const targetFormat =
-    apiFormat === "responses"
-      ? FORMATS.OPENAI_RESPONSES
-      : modelTargetFormat ||
-        customModelTargetFormat ||
-        getTargetFormat(provider, providerSpecificData);
+    modelTargetFormat || customModelTargetFormat || getTargetFormat(provider, providerSpecificData);
   return { alias, targetFormat };
 }
 
-test("apiFormat='responses' short-circuits to OPENAI_RESPONSES (alias still resolved)", () => {
+test("Responses client honors a Chat-compatible upstream target", () => {
+  const provider = "openai-compatible-chat-regression";
+  const r = resolveChatCoreTargetFormat({
+    provider,
+    resolvedModel: "custom-chat-model",
+    apiFormat: "responses",
+    customModelTargetFormat: undefined,
+    providerSpecificData: { apiType: "chat" },
+  });
+  assert.equal(r.targetFormat, FORMATS.OPENAI);
+  assert.equal(r.alias, provider);
+});
+
+test("Responses client preserves a provider configured for native Responses", () => {
+  const provider = "openai-compatible-responses-regression";
+  const r = resolveChatCoreTargetFormat({
+    provider,
+    resolvedModel: "custom-responses-model",
+    apiFormat: "responses",
+    customModelTargetFormat: undefined,
+    providerSpecificData: { apiType: "responses" },
+  });
+  assert.equal(r.targetFormat, FORMATS.OPENAI_RESPONSES);
+});
+
+test("model registry native Responses target overrides a Chat provider default", () => {
+  const model = "gpt-5.6-sol";
+  assert.equal(getModelTargetFormat("openai", model), FORMATS.OPENAI_RESPONSES);
+
   const r = resolveChatCoreTargetFormat({
     provider: "openai",
-    resolvedModel: "gpt-4o",
+    resolvedModel: model,
     apiFormat: "responses",
-    sourceFormat: FORMATS.OPENAI,
     customModelTargetFormat: undefined,
     providerSpecificData: undefined,
   });
   assert.equal(r.targetFormat, FORMATS.OPENAI_RESPONSES);
-  assert.equal(r.alias, PROVIDER_ID_TO_ALIAS["openai"] || "openai");
 });
 
-test("delegates byte-identically for a normal model (no apiFormat / no custom override)", () => {
+test("delegates byte-identically for a normal model without a custom override", () => {
   const r = resolveChatCoreTargetFormat({
     provider: "openai",
     resolvedModel: "gpt-4o",
     apiFormat: undefined,
-    sourceFormat: FORMATS.OPENAI,
     customModelTargetFormat: undefined,
     providerSpecificData: undefined,
   });
-  assert.deepEqual(r, expected("openai", "gpt-4o", undefined, undefined, undefined));
+  assert.deepEqual(r, expected("openai", "gpt-4o", undefined, undefined));
 });
 
 test("customModelTargetFormat is used when the model has no registry target format", () => {
   const customModel = "totally-unknown-custom-model-xyz";
-  // precondition: the registry has no target format for this unknown model
   assert.ok(!getModelTargetFormat(PROVIDER_ID_TO_ALIAS["openai"] || "openai", customModel));
   const r = resolveChatCoreTargetFormat({
     provider: "openai",
     resolvedModel: customModel,
-    apiFormat: undefined,
-    sourceFormat: FORMATS.OPENAI,
+    apiFormat: "responses",
     customModelTargetFormat: "claude",
     providerSpecificData: undefined,
   });
@@ -76,36 +89,25 @@ test("customModelTargetFormat is used when the model has no registry target form
 test("falls back to getTargetFormat(provider) when neither model nor custom format apply", () => {
   const customModel = "totally-unknown-custom-model-xyz";
   const r = resolveChatCoreTargetFormat({
-    provider: "openai",
+    provider: "openai-compatible-chat-regression",
     resolvedModel: customModel,
-    apiFormat: undefined,
-    sourceFormat: FORMATS.OPENAI,
+    apiFormat: "responses",
     customModelTargetFormat: undefined,
-    providerSpecificData: undefined,
+    providerSpecificData: { apiType: "chat" },
   });
-  assert.equal(r.targetFormat, getTargetFormat("openai", undefined));
+  assert.equal(
+    r.targetFormat,
+    getTargetFormat("openai-compatible-chat-regression", { apiType: "chat" })
+  );
 });
 
-test("AgentRouter explicit connection protocol overrides the inferred inbound protocol", () => {
-  const r = resolveChatCoreTargetFormat({
-    provider: "agentrouter",
-    resolvedModel: "gpt-5.6-sol",
-    apiFormat: undefined,
-    sourceFormat: FORMATS.OPENAI_RESPONSES,
-    customModelTargetFormat: undefined,
-    providerSpecificData: { targetFormat: FORMATS.CLAUDE },
-  });
-  assert.equal(r.targetFormat, FORMATS.CLAUDE);
-});
-
-test("unmapped provider → alias falls back to the provider id", () => {
+test("unmapped provider alias falls back to the provider id", () => {
   const r = resolveChatCoreTargetFormat({
     provider: "some-unmapped-provider",
     resolvedModel: "x",
     apiFormat: "responses",
-    sourceFormat: FORMATS.OPENAI,
     customModelTargetFormat: undefined,
     providerSpecificData: undefined,
   });
   assert.equal(r.alias, "some-unmapped-provider");
-});
+});
\ No newline at end of file