deprecate gemini 2 flash and lite (#3581)

* deprecate gemini 2 flash and lite

* fix test

src/backend/drivers/ai-chat/providers/gemini/GeminiChatProvider.integration.test.ts

@@ -20,7 +20,7 @@
 /**
  * Integration test for the Gemini chat provider.
  *
- * Hits the real Google Gemini API with `gemini-2.0-flash-lite` (one
+ * Hits the real Google Gemini API with `gemini-2.5-flash-lite` (one
  * of the cheapest variants). Skipped when `PUTER_TEST_AI_GEMINI_API_KEY`
  * is unset.
  */
@@ -40,14 +40,14 @@ const ENV_VAR = 'PUTER_TEST_AI_GEMINI_API_KEY';
 describe.skipIf(skipUnlessEnv(ENV_VAR))(
     'GeminiChatProvider (integration)',
     () => {
-        it('returns a non-empty completion from gemini-2.0-flash-lite', { timeout: INTEGRATION_TEST_TIMEOUT_MS }, async () => {
+        it('returns a non-empty completion from gemini-2.5-flash-lite', { timeout: INTEGRATION_TEST_TIMEOUT_MS }, async () => {
             const provider = new GeminiChatProvider(makeMeteringStub(), {
                 apiKey: optionalEnv(ENV_VAR)!,
             });
 
             const result = await withTestActor(() =>
                 provider.complete({
-                    model: 'gemini-2.0-flash-lite',
+                    model: 'gemini-2.5-flash-lite',
                     messages: [
                         { role: 'user', content: 'Say hi in one word.' },
                     ],

src/backend/drivers/ai-chat/providers/gemini/GeminiChatProvider.test.ts

@@ -400,46 +400,51 @@ describe('GeminiChatProvider.complete non-stream output', () => {
     });
 
     it('bills cached tokens at the input rate when the model prices no cache read', async () => {
-        // gemini-2.0-flash-lite's catalogue entry has no cached_tokens rate.
-        // Cached tokens are subtracted out of prompt_tokens, so pricing them
-        // at zero bills them nowhere.
+        // Every shipped entry currently prices cache reads, so drop the rate
+        // off one for this call. Cached tokens are subtracted out of
+        // prompt_tokens, so pricing them at zero bills them nowhere.
         const lite = GEMINI_MODELS.find(
-            (m) => m.id === 'gemini-2.0-flash-lite',
+            (m) => m.id === 'gemini-3.1-flash-lite',
         )!;
-        expect(lite.costs.cached_tokens).toBeUndefined();
+        const cachedRate = lite.costs.cached_tokens;
+        delete lite.costs.cached_tokens;
 
-        const { provider } = makeProvider();
-        createMock.mockResolvedValueOnce({
-            choices: [
-                {
-                    message: { content: 'cached', role: 'assistant' },
-                    finish_reason: 'stop',
+        try {
+            const { provider } = makeProvider();
+            createMock.mockResolvedValueOnce({
+                choices: [
+                    {
+                        message: { content: 'cached', role: 'assistant' },
+                        finish_reason: 'stop',
+                    },
+                ],
+                usage: {
+                    prompt_tokens: 3000,
+                    completion_tokens: 40,
+                    prompt_tokens_details: { cached_tokens: 2900 },
                 },
-            ],
-            usage: {
-                prompt_tokens: 3000,
-                completion_tokens: 40,
-                prompt_tokens_details: { cached_tokens: 2900 },
-            },
-        });
+            });
 
-        await withTestActor(() =>
-            provider.complete({
-                model: 'gemini-2.0-flash-lite',
-                messages: [{ role: 'user', content: 'hi' }],
-            }),
-        );
+            await withTestActor(() =>
+                provider.complete({
+                    model: 'gemini-3.1-flash-lite',
+                    messages: [{ role: 'user', content: 'hi' }],
+                }),
+            );
 
-        const [, , , overrides] = recordSpy.mock.calls[0]!;
-        const inputRate = Number(lite.costs.prompt_tokens);
-        expect(overrides).toMatchObject({
-            prompt_tokens: (3000 - 2900) * inputRate,
-            completion_tokens: 40 * Number(lite.costs.completion_tokens),
-            cached_tokens: 2900 * inputRate,
-        });
-        expect(
-            (overrides as Record<string, number>).cached_tokens,
-        ).toBeGreaterThan(0);
+            const [, , , overrides] = recordSpy.mock.calls[0]!;
+            const inputRate = Number(lite.costs.prompt_tokens);
+            expect(overrides).toMatchObject({
+                prompt_tokens: (3000 - 2900) * inputRate,
+                completion_tokens: 40 * Number(lite.costs.completion_tokens),
+                cached_tokens: 2900 * inputRate,
+            });
+            expect(
+                (overrides as Record<string, number>).cached_tokens,
+            ).toBeGreaterThan(0);
+        } finally {
+            lite.costs.cached_tokens = cachedRate;
+        }
     });
 
     it('zeroes cached_tokens when prompt_tokens_details is missing', async () => {
@@ -643,9 +648,9 @@ describe('GeminiChatProvider.complete grounding request metering', () => {
         // generation; without its own rate the fee fell through to the input
         // token rate, which is several orders of magnitude below list.
         const lite = GEMINI_MODELS.find(
-            (m) => m.id === 'gemini-2.0-flash-lite',
+            (m) => m.id === 'gemini-3.1-flash-lite',
         )!;
-        expect(lite.costs.grounding_requests).toBe(3_500_000);
+        expect(lite.costs.grounding_requests).toBe(1_400_000);
 
         const { provider } = makeProvider();
         createMock.mockResolvedValueOnce({
@@ -666,7 +671,7 @@ describe('GeminiChatProvider.complete grounding request metering', () => {
 
         await withTestActor(() =>
             provider.complete({
-                model: 'gemini-2.0-flash-lite',
+                model: 'gemini-3.1-flash-lite',
                 messages: [{ role: 'user', content: 'search for foo' }],
             }),
         );

src/backend/drivers/ai-chat/providers/gemini/models.ts

@@ -49,59 +49,6 @@ export const GEMINI_MODELS: IChatModel[] = [
             grounding_requests: 1_400_000,
         },
     },
-    {
-        puterId: 'google:google/gemini-2.0-flash',
-        id: 'gemini-2.0-flash',
-        modalities: {
-            input: ['text', 'image', 'audio', 'video', 'pdf'],
-            output: ['text'],
-        },
-        open_weights: false,
-        tool_call: true,
-        knowledge: '2024-06',
-        release_date: '2024-12-11',
-        name: 'Gemini 2.0 Flash',
-        aliases: ['google/gemini-2.0-flash'],
-        context: 131072,
-        costs_currency: 'usd-cents',
-        input_cost_key: 'prompt_tokens',
-        output_cost_key: 'completion_tokens',
-        costs: {
-            tokens: 1_000_000,
-            prompt_tokens: 10,
-            completion_tokens: 40,
-            cached_tokens: 3,
-            // Gemini 2.x grounding is $35 / 1,000 requests
-            grounding_requests: 3_500_000,
-        },
-        max_tokens: 8192,
-    },
-    {
-        puterId: 'google:google/gemini-2.0-flash-lite',
-        id: 'gemini-2.0-flash-lite',
-        modalities: {
-            input: ['text', 'image', 'audio', 'video', 'pdf'],
-            output: ['text'],
-        },
-        open_weights: false,
-        tool_call: true,
-        knowledge: '2024-06',
-        release_date: '2024-12-11',
-        name: 'Gemini 2.0 Flash-Lite',
-        aliases: ['google/gemini-2.0-flash-lite'],
-        context: 1_048_576,
-        costs_currency: 'usd-cents',
-        input_cost_key: 'prompt_tokens',
-        output_cost_key: 'completion_tokens',
-        costs: {
-            tokens: 1_000_000,
-            prompt_tokens: 8,
-            completion_tokens: 30,
-            // Gemini 2.x grounding is $35 / 1,000 requests
-            grounding_requests: 3_500_000,
-        },
-        max_tokens: 8192,
-    },
     {
         puterId: 'google:google/gemini-2.5-flash',
         id: 'gemini-2.5-flash',