feat(server): adapt gemini3.1 preview (#14583)

#### PR Dependency Tree


* **PR #14583** 👈

This tree was auto-generated by
[Charcoal](https://github.com/danerwilliams/charcoal)

<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit

* **New Features**
* Added Gemini 3.1 Pro Preview support (text, image, audio) and new
GPT‑5 variants as defaults; centralized persistent telemetry state for
more reliable client identity.

* **UX**
  * Improved model submenu placement in chat preferences.
* More robust mindmap parsing, preview, regeneration and replace
behavior.

* **Chores**
  * Bumped AI SDK and related dependencies.

* **Tests**
  * Expanded/updated tests and increased timeouts for flaky flows.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->

packages/backend/server/src/__tests__/copilot/copilot-provider.spec.ts

@@ -118,7 +118,6 @@ test.serial.before(async t => {
         enabled: true,
         scenarios: {
           image: 'flux-1/schnell',
-          rerank: 'gpt-5-mini',
           complex_text_generation: 'gpt-5-mini',
           coding: 'gpt-5-mini',
           quick_decision_making: 'gpt-5-mini',
@@ -931,8 +930,8 @@ test(
       t.log('Rerank scores:', scores);
       t.is(
         scores.filter(s => s > 0.5).length,
-        4,
-        'should have 4 related chunks'
+        5,
+        'should have 5 related chunks'
       );
     });
   }

packages/backend/server/src/__tests__/copilot/native-adapter.spec.ts

@@ -18,7 +18,7 @@ test('NativeProviderAdapter streamText should append citation footnotes', async
   const adapter = new NativeProviderAdapter(mockDispatch, {}, 3);
   const chunks: string[] = [];
   for await (const chunk of adapter.streamText({
-    model: 'gpt-4.1',
+    model: 'gpt-5-mini',
     stream: true,
     messages: [{ role: 'user', content: [{ type: 'text', text: 'hi' }] }],
   })) {
@@ -36,7 +36,7 @@ test('NativeProviderAdapter streamObject should append citation footnotes', asyn
   const adapter = new NativeProviderAdapter(mockDispatch, {}, 3);
   const chunks = [];
   for await (const chunk of adapter.streamObject({
-    model: 'gpt-4.1',
+    model: 'gpt-5-mini',
     stream: true,
     messages: [{ role: 'user', content: [{ type: 'text', text: 'hi' }] }],
   })) {
@@ -91,7 +91,7 @@ test('NativeProviderAdapter streamObject should append fallback attachment footn
   const adapter = new NativeProviderAdapter(dispatch, {}, 3);
   const chunks = [];
   for await (const chunk of adapter.streamObject({
-    model: 'gpt-4.1',
+    model: 'gpt-5-mini',
     stream: true,
     messages: [{ role: 'user', content: [{ type: 'text', text: 'hi' }] }],
   })) {
@@ -148,7 +148,7 @@ test('NativeProviderAdapter streamObject should map tool and text events', async
 
   const events = [];
   for await (const event of adapter.streamObject({
-    model: 'gpt-4.1',
+    model: 'gpt-5-mini',
     stream: true,
     messages: [{ role: 'user', content: [{ type: 'text', text: 'read' }] }],
   })) {
@@ -169,7 +169,7 @@ test('NativeProviderAdapter streamObject should map tool and text events', async
 
 test('buildNativeRequest should include rust middleware from profile', async t => {
   const { request } = await buildNativeRequest({
-    model: 'gpt-4.1',
+    model: 'gpt-5-mini',
     messages: [{ role: 'user', content: 'hello' }],
     tools: {},
     middleware: {
@@ -195,7 +195,7 @@ test('NativeProviderAdapter streamText should skip citation footnotes when disab
   });
   const chunks: string[] = [];
   for await (const chunk of adapter.streamText({
-    model: 'gpt-4.1',
+    model: 'gpt-5-mini',
     stream: true,
     messages: [{ role: 'user', content: [{ type: 'text', text: 'hi' }] }],
   })) {

packages/backend/server/src/__tests__/copilot/provider-native.spec.ts

@@ -1,7 +1,9 @@
 import test from 'ava';
 
 import { ProviderMiddlewareConfig } from '../../plugins/copilot/config';
+import { normalizeOpenAIOptionsForModel } from '../../plugins/copilot/providers/openai';
 import { CopilotProvider } from '../../plugins/copilot/providers/provider';
+import { normalizeRerankModel } from '../../plugins/copilot/providers/rerank';
 import {
   CopilotProviderType,
   ModelInputType,
@@ -12,7 +14,7 @@ class TestOpenAIProvider extends CopilotProvider<{ apiKey: string }> {
   readonly type = CopilotProviderType.OpenAI;
   readonly models = [
     {
-      id: 'gpt-4.1',
+      id: 'gpt-5-mini',
       capabilities: [
         {
           input: [ModelInputType.Text],
@@ -36,7 +38,7 @@ class TestOpenAIProvider extends CopilotProvider<{ apiKey: string }> {
   }
 
   exposeMetricLabels() {
-    return this.metricLabels('gpt-4.1');
+    return this.metricLabels('gpt-5-mini');
   }
 
   exposeMiddleware() {
@@ -97,3 +99,41 @@ test('getActiveProviderMiddleware should merge defaults with profile override',
     'thinking_format',
   ]);
 });
+
+test('normalizeOpenAIOptionsForModel should drop sampling knobs for gpt-5.2', t => {
+  t.deepEqual(
+    normalizeOpenAIOptionsForModel(
+      {
+        temperature: 0.7,
+        topP: 0.8,
+        presencePenalty: 0.2,
+        frequencyPenalty: 0.1,
+        maxTokens: 128,
+      },
+      'gpt-5.4'
+    ),
+    { maxTokens: 128 }
+  );
+});
+
+test('normalizeOpenAIOptionsForModel should keep options for gpt-4.1', t => {
+  t.deepEqual(
+    normalizeOpenAIOptionsForModel(
+      { temperature: 0.7, topP: 0.8, maxTokens: 128 },
+      'gpt-4.1'
+    ),
+    { temperature: 0.7, topP: 0.8, maxTokens: 128 }
+  );
+});
+
+test('normalizeOpenAIRerankModel should keep supported rerank models', t => {
+  t.is(normalizeRerankModel('gpt-4.1'), 'gpt-4.1');
+  t.is(normalizeRerankModel('gpt-4.1-mini'), 'gpt-4.1-mini');
+  t.is(normalizeRerankModel('gpt-5.2'), 'gpt-5.2');
+});
+
+test('normalizeOpenAIRerankModel should fall back for unsupported models', t => {
+  t.is(normalizeRerankModel('gpt-5-mini'), 'gpt-5.2');
+  t.is(normalizeRerankModel('gemini-2.5-flash'), 'gpt-5.2');
+  t.is(normalizeRerankModel(undefined), 'gpt-5.2');
+});

packages/backend/server/src/__tests__/copilot/provider-registry.spec.ts

@@ -88,11 +88,11 @@ test('resolveModel should support explicit provider prefix and keep slash models
 
   const prefixed = resolveModel({
     registry,
-    modelId: 'openai-main/gpt-4.1',
+    modelId: 'openai-main/gpt-5-mini',
   });
   t.deepEqual(prefixed, {
-    rawModelId: 'openai-main/gpt-4.1',
-    modelId: 'gpt-4.1',
+    rawModelId: 'openai-main/gpt-5-mini',
+    modelId: 'gpt-5-mini',
     explicitProviderId: 'openai-main',
     candidateProviderIds: ['openai-main'],
   });
@@ -154,12 +154,15 @@ test('stripProviderPrefix should only strip matched provider prefix', t => {
   });
 
   t.is(
-    stripProviderPrefix(registry, 'openai-main', 'openai-main/gpt-4.1'),
-    'gpt-4.1'
+    stripProviderPrefix(registry, 'openai-main', 'openai-main/gpt-5-mini'),
+    'gpt-5-mini'
   );
   t.is(
-    stripProviderPrefix(registry, 'openai-main', 'another-main/gpt-4.1'),
-    'another-main/gpt-4.1'
+    stripProviderPrefix(registry, 'openai-main', 'another-main/gpt-5-mini'),
+    'another-main/gpt-5-mini'
+  );
+  t.is(
+    stripProviderPrefix(registry, 'openai-main', 'gpt-5-mini'),
+    'gpt-5-mini'
   );
-  t.is(stripProviderPrefix(registry, 'openai-main', 'gpt-4.1'), 'gpt-4.1');
 });

packages/backend/server/src/__tests__/copilot/tool-call-loop.spec.ts

@@ -116,7 +116,7 @@ test('ToolCallLoop should execute tool call and continue to next round', async t
 
   const events: NativeLlmStreamEvent[] = [];
   for await (const event of loop.run({
-    model: 'gpt-4.1',
+    model: 'gpt-5-mini',
     stream: true,
     messages: [{ role: 'user', content: [{ type: 'text', text: 'read doc' }] }],
   })) {

packages/backend/server/src/__tests__/mocks/copilot.mock.ts

@@ -39,33 +39,6 @@ export class MockCopilotProvider extends OpenAIProvider {
         },
       ],
     },
-    {
-      id: 'gpt-4o',
-      capabilities: [
-        {
-          input: [ModelInputType.Text, ModelInputType.Image],
-          output: [ModelOutputType.Text, ModelOutputType.Object],
-        },
-      ],
-    },
-    {
-      id: 'gpt-4o-2024-08-06',
-      capabilities: [
-        {
-          input: [ModelInputType.Text, ModelInputType.Image],
-          output: [ModelOutputType.Text, ModelOutputType.Object],
-        },
-      ],
-    },
-    {
-      id: 'gpt-4.1-2025-04-14',
-      capabilities: [
-        {
-          input: [ModelInputType.Text, ModelInputType.Image],
-          output: [ModelOutputType.Text, ModelOutputType.Object],
-        },
-      ],
-    },
     {
       id: 'gpt-5',
       capabilities: [
@@ -97,6 +70,19 @@ export class MockCopilotProvider extends OpenAIProvider {
         },
       ],
     },
+    {
+      id: 'gpt-5-nano',
+      capabilities: [
+        {
+          input: [ModelInputType.Text, ModelInputType.Image],
+          output: [
+            ModelOutputType.Text,
+            ModelOutputType.Object,
+            ModelOutputType.Structured,
+          ],
+        },
+      ],
+    },
     {
       id: 'gpt-image-1',
       capabilities: [
@@ -133,6 +119,23 @@ export class MockCopilotProvider extends OpenAIProvider {
         },
       ],
     },
+    {
+      id: 'gemini-3.1-pro-preview',
+      capabilities: [
+        {
+          input: [
+            ModelInputType.Text,
+            ModelInputType.Image,
+            ModelInputType.Audio,
+          ],
+          output: [
+            ModelOutputType.Text,
+            ModelOutputType.Object,
+            ModelOutputType.Structured,
+          ],
+        },
+      ],
+    },
   ];
 
   override async text(