fix(usage): parse CREDIT_LIMIT rows from z.ai coding-plan quota API (#11378)

Merged via consolidated batch validation. Z.ai's quota API now returns CREDIT_LIMIT rows for GLM Coding Plan subscription keys instead of TOKENS_LIMIT, breaking the dashboard quota card. Own test passes.

changelog.d/fixes/glm-credit-limit-quota.md

@@ -0,0 +1 @@
+- **fix(usage):** z.ai/GLM coding-plan subscription keys now render their quota cards again, with absolute credits. Z.ai's `/api/monitor/usage/quota/limit` switched these keys from `TOKENS_LIMIT` to `CREDIT_LIMIT` rows (same `unit`/`number` semantics: unit=3/number=5 → 5-hour window, unit=6/number=1 → weekly), and the parser only matched `TOKENS_LIMIT`/`TIME_LIMIT`, so both rows were dropped and the subscription card rendered empty. `CREDIT_LIMIT` is now accepted alongside `TOKENS_LIMIT`, and when the row carries absolute credit fields (`usage`/`currentValue`/`remaining`) they are preferred over the percent-only scale, so the card shows `3341 / 28000` like z.ai's own dashboard instead of `11 / 100`

open-sse/services/usage/glm.ts

@@ -155,15 +155,30 @@ export async function getGlmUsage(apiKey: string, providerSpecificData?: Record<
     const resetMs = toNumber(src.nextResetTime, 0);
     const resetAt = resetMs > 0 ? new Date(resetMs).toISOString() : null;
 
-    if (type === "TOKENS_LIMIT") {
+    // Z.ai coding-plan keys (CREDIT-based, e.g. GLM Coding Max/Lite) report
+    // CREDIT_LIMIT rows with the same unit/number semantics as TOKENS_LIMIT
+    // (unit=3/number=5 → 5-hour window, unit=6/number=1 → weekly). Without
+    // this branch every CREDIT_LIMIT row is dropped and the quota card
+    // renders empty for subscription keys.
+    if (type === "TOKENS_LIMIT" || type === "CREDIT_LIMIT") {
       const quotaName = getGlmTokenQuotaName(src, quotas);
       const usedPercent = toPercentage(src.percentage);
       const remaining = Math.max(0, 100 - usedPercent);
 
+      // CREDIT_LIMIT rows (z.ai coding-plan keys) carry absolute credits on
+      // top of the percentage: usage = window total, currentValue = consumed,
+      // remaining = credits left. Prefer them so the quota card renders
+      // "3341 / 28000" like z.ai's own dashboard instead of a percent-only
+      // scale. TOKENS_LIMIT rows without absolute fields keep the percent path.
+      const totalCredits = toNumber(src.usage, 0);
+      const usedCredits = totalCredits > 0 ? toNumber(src.currentValue, usedPercent) : usedPercent;
+      const remainingCredits = totalCredits > 0 ? toNumber(src.remaining, remaining) : remaining;
+      const total = totalCredits > 0 ? totalCredits : 100;
+
       quotas[quotaName] = {
-        used: usedPercent,
-        total: 100,
-        remaining,
+        used: usedCredits,
+        total,
+        remaining: remainingCredits,
         remainingPercentage: remaining,
         resetAt,
         displayName: getGlmQuotaDisplayName(quotaName),

tests/unit/glm-team-quota.test.ts

@@ -317,3 +317,96 @@ describe("getGlmUsage team quota parsing", () => {
     }
   });
 });
+
+describe("getGlmUsage CREDIT_LIMIT (coding-plan subscription keys)", () => {
+  // Real-world response from https://api.z.ai/api/monitor/usage/quota/limit
+  // for a GLM Coding Max subscription key (2026-08): limits use CREDIT_LIMIT
+  // instead of TOKENS_LIMIT, with identical unit/number semantics plus
+  // absolute credit fields (usage/currentValue/remaining).
+  const CREDIT_LIMIT_RESPONSE = {
+    code: 200,
+    msg: "Operation successful",
+    data: {
+      limits: [
+        {
+          type: "CREDIT_LIMIT",
+          unit: 3,
+          number: 5,
+          usage: 28000,
+          currentValue: 3341,
+          remaining: 24658,
+          percentage: 11,
+          nextResetTime: 1787563232239,
+        },
+        {
+          type: "CREDIT_LIMIT",
+          unit: 6,
+          number: 1,
+          usage: 140000,
+          currentValue: 25224,
+          remaining: 114775,
+          percentage: 18,
+          nextResetTime: 1788077327998,
+        },
+      ],
+      level: "max",
+    },
+    success: true,
+  };
+
+  it("maps CREDIT_LIMIT rows to session/weekly quotas with absolute credits", async () => {
+    const originalFetch = globalThis.fetch;
+    globalThis.fetch = async () =>
+      new Response(JSON.stringify(CREDIT_LIMIT_RESPONSE), { status: 200 });
+
+    try {
+      const usage = await getGlmUsage("zai-subscription-key");
+
+      assert.equal(usage.plan, "Max");
+      assert.ok(usage.quotas.session, "5-hour window quota should render");
+      assert.ok(usage.quotas.weekly, "weekly quota should render");
+      // Absolute credits — matches z.ai's own dashboard ("4.1K / 140K" style).
+      assert.equal(usage.quotas.session.used, 3341);
+      assert.equal(usage.quotas.session.total, 28000);
+      assert.equal(usage.quotas.session.remaining, 24658);
+      assert.equal(usage.quotas.weekly.used, 25224);
+      assert.equal(usage.quotas.weekly.total, 140000);
+      assert.equal(usage.quotas.weekly.remaining, 114775);
+      // Percentages stay derived from the upstream percentage field.
+      assert.equal(usage.quotas.session.remainingPercentage, 89);
+      assert.equal(usage.quotas.weekly.remainingPercentage, 82);
+      assert.equal(usage.quotas.session.displayName, "5 Hours Quota");
+      assert.equal(usage.quotas.weekly.displayName, "Weekly Quota");
+      assert.equal(usage.quotas.session.resetAt, new Date(1787563232239).toISOString());
+    } finally {
+      globalThis.fetch = originalFetch;
+    }
+  });
+
+  it("falls back to the percent scale when a CREDIT_LIMIT row has no absolute fields", async () => {
+    const originalFetch = globalThis.fetch;
+    globalThis.fetch = async () =>
+      new Response(
+        JSON.stringify({
+          code: 200,
+          success: true,
+          data: {
+            limits: [{ type: "CREDIT_LIMIT", unit: 3, number: 5, percentage: 40 }],
+            level: "lite",
+          },
+        }),
+        { status: 200 }
+      );
+
+    try {
+      const usage = await getGlmUsage("zai-key");
+
+      assert.equal(usage.quotas.session.used, 40);
+      assert.equal(usage.quotas.session.total, 100);
+      assert.equal(usage.quotas.session.remaining, 60);
+      assert.equal(usage.quotas.session.remainingPercentage, 60);
+    } finally {
+      globalThis.fetch = originalFetch;
+    }
+  });
+});