Models availability e2e test fix - new default model gpt-5.6-sol/medium (agentclientprotocol#335)

Author: Rizzen

src/__tests__/CodexACPAgent/e2e/acp-e2e-models-availability.test.ts

@@ -2,6 +2,8 @@ import {afterEach, beforeEach, expect, it} from "vitest";
 import {createAuthenticatedFixture, describeE2E, type SpawnedAgentFixture,} from "./acp-e2e-test-utils";
 import {ModelId} from "../../../ModelId";
 
+const DEFAULT_MODEL_ID = ModelId.create("gpt-5.6-sol", "medium")
+
 describeE2E("Models availability", () => {
     let fixture: SpawnedAgentFixture;
 
@@ -15,22 +17,7 @@ describeE2E("Models availability", () => {
 
     it(`default model is available`, async () => {
         const session = await fixture.createSession();
-        const models = session.models;
-        const availableModelIds = models?.availableModels?.map(m => m.modelId) ?? [];
-        expect(availableModelIds.length).toBeGreaterThan(0);
-
-        // Codex's advertised catalog changes as it is upgraded, so assert the
-        // invariant that survives those bumps instead of pinning a specific
-        // model id: the session's current (default) model must be one of the
-        // advertised models. Compare on the base model id because
-        // availableModels enumerate model x reasoning-effort pairs while the
-        // current model may carry an effort (e.g. "none") that is not itself
-        // enumerated.
-        const currentModelId = models?.currentModelId;
-        expect(currentModelId).toBeDefined();
-
-        const currentBaseModel = ModelId.fromString(currentModelId!).model;
-        const availableBaseModels = availableModelIds.map(id => ModelId.fromString(id).model);
-        expect(availableBaseModels).toContain(currentBaseModel);
+        const models = session.models?.availableModels?.map(m => m.modelId);
+        expect(models).toContain(DEFAULT_MODEL_ID.toString())
     });
 });