fix: normalize QA model refs for parity gates

.github/workflows/openclaw-release-checks.yml

@@ -62,7 +62,7 @@ env:
   FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
   NODE_VERSION: "24.x"
   PNPM_VERSION: "10.33.0"
-  OPENCLAW_CI_OPENAI_MODEL: ${{ vars.OPENCLAW_CI_OPENAI_MODEL }}
+  OPENCLAW_CI_OPENAI_MODEL: ${{ vars.OPENCLAW_CI_OPENAI_MODEL || 'openai/gpt-5.5' }}
 
 jobs:
   resolve_target:

.github/workflows/parity-gate.yml

@@ -42,7 +42,7 @@ jobs:
       # followthrough gate that expects a fast post-approval read within a 30s
       # agent.wait timeout.
       QA_PARITY_CONCURRENCY: "1"
-      OPENCLAW_CI_OPENAI_MODEL: ${{ vars.OPENCLAW_CI_OPENAI_MODEL }}
+      OPENCLAW_CI_OPENAI_MODEL: ${{ vars.OPENCLAW_CI_OPENAI_MODEL || 'openai/gpt-5.5' }}
       OPENCLAW_QA_TRANSPORT_READY_TIMEOUT_MS: "180000"
       OPENAI_API_KEY: ""
       ANTHROPIC_API_KEY: ""

.github/workflows/qa-live-transports-convex.yml

@@ -44,7 +44,7 @@ env:
   FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: "true"
   NODE_VERSION: "24.x"
   PNPM_VERSION: "10.33.0"
-  OPENCLAW_CI_OPENAI_MODEL: ${{ vars.OPENCLAW_CI_OPENAI_MODEL }}
+  OPENCLAW_CI_OPENAI_MODEL: ${{ vars.OPENCLAW_CI_OPENAI_MODEL || 'openai/gpt-5.5' }}
   OPENCLAW_BUILD_PRIVATE_QA: "1"
   OPENCLAW_ENABLE_PRIVATE_QA_CLI: "1"
 

extensions/qa-lab/src/cli.runtime.test.ts

@@ -206,6 +206,27 @@ describe("qa cli runtime", () => {
     });
   });
 
+  it("drops blank suite model refs so provider defaults apply", async () => {
+    await runQaSuiteCommand({
+      repoRoot: "/tmp/openclaw-repo",
+      providerMode: "mock-openai",
+      primaryModel: " ",
+      alternateModel: "",
+      scenarioIds: ["thread-memory-isolation"],
+    });
+
+    expect(runQaSuiteFromRuntime).toHaveBeenCalledWith({
+      repoRoot: path.resolve("/tmp/openclaw-repo"),
+      outputDir: undefined,
+      transportId: "qa-channel",
+      providerMode: "mock-openai",
+      primaryModel: undefined,
+      alternateModel: undefined,
+      fastMode: undefined,
+      scenarioIds: ["thread-memory-isolation"],
+    });
+  });
+
   it("resolves telegram qa repo-root-relative paths before dispatching", async () => {
     await runQaTelegramCommand({
       repoRoot: "/tmp/openclaw-repo",

extensions/qa-lab/src/cli.runtime.ts

@@ -127,6 +127,11 @@ function parseQaPositiveIntegerOption(label: string, value: number | undefined)
   return Math.floor(value);
 }
 
+function normalizeQaOptionalModelRef(input: string | undefined) {
+  const model = input?.trim();
+  return model && model.length > 0 ? model : undefined;
+}
+
 async function readQaFailedScenarioCountFromSummary(summaryPath: string) {
   let summaryText: string;
   try {
@@ -488,6 +493,8 @@ export async function runQaSuiteCommand(opts: {
   }
   const providerMode = normalizeQaProviderMode(opts.providerMode);
   const claudeCliAuthMode = parseQaCliBackendAuthMode(opts.cliAuthMode);
+  const primaryModel = normalizeQaOptionalModelRef(opts.primaryModel);
+  const alternateModel = normalizeQaOptionalModelRef(opts.alternateModel);
   if (opts.preflight === true && runner !== "host") {
     throw new Error("--preflight requires --runner host.");
   }
@@ -510,8 +517,8 @@ export async function runQaSuiteCommand(opts: {
       outputDir: resolveRepoRelativeOutputDir(repoRoot, opts.outputDir),
       transportId,
       providerMode,
-      primaryModel: opts.primaryModel,
-      alternateModel: opts.alternateModel,
+      primaryModel,
+      alternateModel,
       fastMode: opts.fastMode,
       ...(thinkingDefault ? { thinkingDefault } : {}),
       allowFailures: true,
@@ -542,8 +549,8 @@ export async function runQaSuiteCommand(opts: {
       repoRoot,
       transportId,
       providerMode,
-      primaryModel: opts.primaryModel,
-      alternateModel: opts.alternateModel,
+      primaryModel,
+      alternateModel,
       allowFailures,
     });
     return;
@@ -554,8 +561,8 @@ export async function runQaSuiteCommand(opts: {
     outputDir: resolveRepoRelativeOutputDir(repoRoot, opts.outputDir),
     transportId,
     providerMode,
-    primaryModel: opts.primaryModel,
-    alternateModel: opts.alternateModel,
+    primaryModel,
+    alternateModel,
     fastMode: opts.fastMode,
     ...(thinkingDefault ? { thinkingDefault } : {}),
     ...(claudeCliAuthMode ? { claudeCliAuthMode } : {}),

extensions/qa-lab/src/live-transports/shared/live-transport-cli.runtime.test.ts

@@ -0,0 +1,20 @@
+import { describe, expect, it } from "vitest";
+import { resolveLiveTransportQaRunOptions } from "./live-transport-cli.runtime.js";
+
+describe("resolveLiveTransportQaRunOptions", () => {
+  it("drops blank model refs so live transports can use provider defaults", () => {
+    expect(
+      resolveLiveTransportQaRunOptions({
+        repoRoot: "/tmp/openclaw-repo",
+        providerMode: "live-frontier",
+        primaryModel: " ",
+        alternateModel: "",
+      }),
+    ).toMatchObject({
+      repoRoot: "/tmp/openclaw-repo",
+      providerMode: "live-frontier",
+      primaryModel: undefined,
+      alternateModel: undefined,
+    });
+  });
+});

extensions/qa-lab/src/live-transports/shared/live-transport-cli.runtime.ts

@@ -5,6 +5,11 @@ import type { QaProviderMode } from "../../run-config.js";
 import { normalizeQaProviderMode } from "../../run-config.js";
 import type { LiveTransportQaCommandOptions } from "./live-transport-cli.js";
 
+function normalizeLiveTransportModelRef(input: string | undefined) {
+  const model = input?.trim();
+  return model && model.length > 0 ? model : undefined;
+}
+
 export function resolveLiveTransportQaRunOptions(
   opts: LiveTransportQaCommandOptions,
 ): LiveTransportQaCommandOptions & {
@@ -21,8 +26,8 @@ export function resolveLiveTransportQaRunOptions(
       opts.providerMode === undefined
         ? DEFAULT_QA_LIVE_PROVIDER_MODE
         : normalizeQaProviderMode(opts.providerMode),
-    primaryModel: opts.primaryModel,
-    alternateModel: opts.alternateModel,
+    primaryModel: normalizeLiveTransportModelRef(opts.primaryModel),
+    alternateModel: normalizeLiveTransportModelRef(opts.alternateModel),
     fastMode: opts.fastMode,
     allowFailures: opts.allowFailures,
     scenarioIds: opts.scenarioIds,

extensions/qa-lab/src/qa-gateway-config.test.ts

@@ -109,6 +109,22 @@ describe("buildQaGatewayConfig", () => {
     expect(cfg.plugins?.allow).toEqual(["acpx", "memory-core"]);
   });
 
+  it("falls back to provider defaults for blank model refs", () => {
+    const cfg = buildQaGatewayConfig({
+      bind: "loopback",
+      gatewayPort: 18789,
+      gatewayToken: "token",
+      providerBaseUrl: "http://127.0.0.1:44080/v1",
+      workspaceDir: "/tmp/qa-workspace",
+      providerMode: "mock-openai",
+      primaryModel: " ",
+      alternateModel: "",
+    });
+
+    expect(getPrimaryModel(cfg.agents?.defaults?.model)).toBe("mock-openai/gpt-5.5");
+    expect(cfg.agents?.defaults?.models).toHaveProperty("mock-openai/gpt-5.5-alt");
+  });
+
   it("can wire AIMock as a separate mock provider lane", () => {
     const cfg = buildQaGatewayConfig({
       bind: "loopback",

extensions/qa-lab/src/qa-gateway-config.ts

@@ -29,6 +29,11 @@ export function mergeQaControlUiAllowedOrigins(extraOrigins?: string[]) {
   return [...new Set([...DEFAULT_QA_CONTROL_UI_ALLOWED_ORIGINS, ...normalizedExtra])];
 }
 
+function normalizeQaGatewayModelRef(input: string | undefined, fallback: string) {
+  const model = input?.trim();
+  return model && model.length > 0 ? model : fallback;
+}
+
 export function buildQaGatewayConfig(params: {
   bind: "loopback" | "lan";
   gatewayPort: number;
@@ -53,9 +58,14 @@ export function buildQaGatewayConfig(params: {
   const providerBaseUrl = params.providerBaseUrl ?? "http://127.0.0.1:44080/v1";
   const providerMode = normalizeQaProviderMode(params.providerMode ?? DEFAULT_QA_PROVIDER_MODE);
   const provider = getQaProvider(providerMode);
-  const primaryModel = params.primaryModel ?? defaultQaModelForMode(providerMode);
-  const alternateModel =
-    params.alternateModel ?? defaultQaModelForMode(providerMode, { alternate: true });
+  const primaryModel = normalizeQaGatewayModelRef(
+    params.primaryModel,
+    defaultQaModelForMode(providerMode),
+  );
+  const alternateModel = normalizeQaGatewayModelRef(
+    params.alternateModel,
+    defaultQaModelForMode(providerMode, { alternate: true }),
+  );
   const modelProviderIds = [primaryModel, alternateModel]
     .map((ref) => splitQaModelRef(ref)?.provider)
     .filter((provider): provider is string => Boolean(provider));

extensions/qa-lab/src/suite.ts

@@ -273,6 +273,11 @@ function createQaSuiteReportNotes(params: {
   return params.transport.createReportNotes(params);
 }
 
+function normalizeQaSuiteModelRef(input: string | undefined, fallback: string) {
+  const model = input?.trim();
+  return model && model.length > 0 ? model : fallback;
+}
+
 export type QaSuiteSummaryJsonParams = {
   scenarios: QaSuiteScenarioResult[];
   startedAt: Date;
@@ -407,8 +412,14 @@ export async function runQaSuite(params?: QaSuiteRunParams): Promise<QaSuiteResu
     params?.providerMode ?? DEFAULT_QA_LIVE_PROVIDER_MODE,
   );
   const transportId = normalizeQaTransportId(params?.transportId);
-  const primaryModel = params?.primaryModel ?? defaultQaModelForMode(providerMode);
-  const alternateModel = params?.alternateModel ?? defaultQaModelForMode(providerMode, true);
+  const primaryModel = normalizeQaSuiteModelRef(
+    params?.primaryModel,
+    defaultQaModelForMode(providerMode),
+  );
+  const alternateModel = normalizeQaSuiteModelRef(
+    params?.alternateModel,
+    defaultQaModelForMode(providerMode, true),
+  );
   const fastMode =
     typeof params?.fastMode === "boolean"
       ? params.fastMode