fix(ai): make model capability defaults forward-compatible (#52388)

packages/ai/src/protocols/openai-responses.ts

@@ -143,14 +143,15 @@ const adapter = {
   restoreHostedToolItem: (item: unknown) => (Schema.is(OpenAIResponsesHostedToolItem)(item) ? item : undefined),
 } satisfies OpenResponses.ProviderAdapter
 
-// GPT-6 Astra, Sol, and Luna accept `configuration_update` only in standard mode (not `reasoning.mode: "pro"` or
-// `-pro` slugs), and never alongside automatic `context_management` compaction.
+// GPT-6 and later default to `configuration_update` support, except in `reasoning.mode: "pro"`
+// or alongside automatic `context_management` compaction.
 const supportsEffortUpdates = (request: LLMRequest) => {
   if (request.providerOptions?.contextManagement !== undefined) return false
   if (Schema.is(Schema.Struct({ mode: Schema.Literal("pro") }))(request.http?.body?.reasoning)) return false
   const override = request.model.compatibility?.supportsEffortUpdates
   if (override !== undefined) return override
-  return /(?:^|\/)gpt-6-(?:astra|sol|luna)$/i.test(request.model.id)
+  const match = /(?:^|\/)gpt-(\d+)(?:\.\d+)?(?:-|$)/i.exec(request.model.id)
+  return match !== null && Number(match[1]) >= 6
 }
 
 const nativeImageToolInput = (tool: ToolDefinition) => {

packages/ai/src/protocols/zai-chat.ts

@@ -41,16 +41,20 @@ const Body = Schema.Struct({
   user_id: Options.fields.userID,
 })
 
+// Tool streaming was introduced in GLM-4.6; later versions inherit support.
+const supportsToolStreaming = (modelID: string) => {
+  const match = /(?:^|\/)glm-(\d+)(?:\.(\d+))?(?:-|$)/i.exec(modelID)
+  return match !== null && (Number(match[1]) > 4 || (Number(match[1]) === 4 && Number(match[2] ?? 0) >= 6))
+}
+
 const fromRequest = Effect.fn("ZAIChat.fromRequest")(function* (request: LLMRequest) {
   const options = yield* ProviderShared.validateWith(Schema.decodeUnknownEffect(Options))(request.providerOptions ?? {})
   const body = yield* OpenAIChat.protocol.body.from(request)
   return {
     ...body,
     thinking: options.thinking,
-    // Tool streaming was introduced in GLM-4.6; older models must not receive the opt-in.
     tool_stream:
-      options.toolStream ??
-      (body.tools?.length && /^glm-(?:4\.[67]|5(?:[.-]|$))/i.test(request.model.id) ? true : undefined),
+      options.toolStream ?? (body.tools?.length && supportsToolStreaming(request.model.id) ? true : undefined),
     do_sample: options.doSample,
     response_format: options.responseFormat,
     request_id: options.requestID,

packages/ai/test/effort-updates.test.ts

@@ -375,11 +375,24 @@ describe("OpenAI Responses effort updates", () => {
     ["openai/gpt-6-sol", true],
     ["gpt-6-luna", true],
     ["openai/gpt-6-luna", true],
-    ["gpt-6-astra-2026-09-01", false],
-    ["gpt-6-sol-pro", false],
-    ["gpt-6-luna-pro", false],
-    ["gpt-6-sol-fast", false],
+    ["gpt-6", true],
+    ["gpt-6.1-sol", true],
+    ["openai/gpt-6.1-sol", true],
+    ["GPT-6.1-SOL", true],
+    ["gpt-6-astra-2026-09-01", true],
+    ["gpt-6-sol-pro", true],
+    ["gpt-6-luna-pro", true],
+    ["gpt-6-sol-fast", true],
+    ["gpt-7", true],
+    ["openai/gpt-7.2-new-family", true],
+    ["gpt-10.1", true],
+    ["gpt-5", false],
     ["gpt-5.6-sol", false],
+    ["gpt-5.10", false],
+    ["not-gpt-6-sol", false],
+    ["gpt-6foo", false],
+    ["gpt-6.x-sol", false],
+    ["future-model", false],
   ] as const) {
     it.effect(`${supported ? "lowers" : "strips"} markers for ${id}`, () =>
       Effect.gen(function* () {

packages/ai/test/provider/zai.test.ts

@@ -54,16 +54,28 @@ it.effect("ZAI Chat retains native thinking fields and scopes tool streaming to
         baseURL: "https://gateway.example/custom",
         auth: Auth.header("x-test-key", "fixture"),
       })
-      for (const modelID of [
-        "glm-4.5",
-        "glm-4.5-air",
-        "glm-4.5-x",
-        "glm-4.6",
-        "glm-4.7-flash",
-        "glm-5.3",
-        "glm-5.3-flash",
-        "future-model",
-      ]) {
+      for (const [modelID, supported] of [
+        ["glm-3.9", false],
+        ["glm-4", false],
+        ["glm-4.5", false],
+        ["glm-4.5-air", false],
+        ["glm-4.5-x", false],
+        ["glm-4.6", true],
+        ["glm-4.7-flash", true],
+        ["glm-4.8", true],
+        ["glm-4.10", true],
+        ["glm-5", true],
+        ["glm-5.3", true],
+        ["glm-5.3-flash", true],
+        ["glm-6", true],
+        ["glm-10.1-new-family", true],
+        ["zai/glm-6.1", true],
+        ["GLM-4.6", true],
+        ["not-glm-6", false],
+        ["glm-4.60foo", false],
+        ["glm-6.x", false],
+        ["future-model", false],
+      ] as const) {
         const result = yield* compileRequest(
           LLM.request({
             model: provider.chat(modelID),
@@ -78,9 +90,7 @@ it.effect("ZAI Chat retains native thinking fields and scopes tool streaming to
           max_tokens: 123,
           thinking: { type: "enabled", clear_thinking: false },
         })
-        expect(result.body.tool_stream).toBe(
-          ["glm-4.6", "glm-4.7-flash", "glm-5.3", "glm-5.3-flash"].includes(modelID) ? true : undefined,
-        )
+        expect(result.body.tool_stream).toBe(supported ? true : undefined)
         expect(result.body.max_completion_tokens).toBeUndefined()
         expect(result.body.store).toBeUndefined()
         expect(result.body.tools).toEqual([
@@ -120,6 +130,30 @@ it.effect("ZAI Chat retains native thinking fields and scopes tool streaming to
   }),
 )
 
+it.effect("ZAI tool streaming defaults require tools and preserve explicit overrides", () =>
+  Effect.gen(function* () {
+    for (const configure of [ZAI.configure, ZAICodingPlan.configure]) {
+      const provider = configure({ apiKey: "fixture" })
+      for (const item of [
+        { id: "glm-6.1", tools: [], toolStream: undefined, expected: undefined },
+        { id: "glm-4.5", tools: [tool], toolStream: true, expected: true },
+        { id: "glm-6.1", tools: [tool], toolStream: false, expected: false },
+        { id: "future-model", tools: [tool], toolStream: true, expected: true },
+      ]) {
+        const result = yield* compileRequest(
+          LLM.request({
+            model: provider.chat(item.id),
+            prompt: "Hello",
+            tools: item.tools,
+            providerOptions: { toolStream: item.toolStream },
+          }),
+        )
+        expect(result.body.tool_stream).toBe(item.expected)
+      }
+    }
+  }),
+)
+
 it.effect("ZAI lowers effort using the selected native API without inventing thinking settings", () =>
   Effect.gen(function* () {
     for (const effort of ["none", "minimal", "low", "medium", "high", "xhigh", "max", "future-effort"]) {