# Can't turn off thinking mode using Gemma4 **URL:** \ **Category:** Gemini API\ **Tags:** thinking\ **Created:** [April 20, 2026, 4:13am UTC](https://discuss.ai.google.dev/t/cant-turn-off-thinking-mode-using-gemma4/141243 "2026-04-20T04:13:22Z")\ **Posts on this page:** 1\ **Showing post:** 1
**Author:** ![Greg\_Obleshchuk](https://d1dlmcr85iqnpo.cloudfront.net/user_avatar/discuss.ai.google.dev/greg_obleshchuk/32/32294_2.png) [@Greg\_Obleshchuk](https://discuss.ai.google.dev/u/Greg_Obleshchuk)\ **Post date:** [April 20, 2026, 4:13am UTC](https://discuss.ai.google.dev/t/cant-turn-off-thinking-mode-using-gemma4/141243/1 "2026-04-20T04:13:22Z")
Hi , I’m calling the gemma4 model both gemma-4-31b-it and gemma-4-26b-a4b-it via rest api and I can’t runoff thinking. this is my url [https://generativelanguage.googleapis.com/v1beta/models/gemma-4-31b-it:streamGenerateContent?key=33333333](https://generativelanguage.googleapis.com/v1beta/models/gemma-4-31b-it:streamGenerateContent?key=33333333) and here is the body { ``` "contents": \[ { "role": "user", "parts": \[ { "text": "Who are you" } \] } \], "generationConfig": { "maxOutputTokens": 4048, "responseMimeType": "application/json", "frequencyPenalty": 0.0, "presencePenalty": 0.0, "temperature": 0.0 , "topP": 1 }, "systemInstruction": { "parts": \[ { "text": "You are a AI assistant." } \] ``` } } here are the results [ ``` { "candidates": \[ { "content": { "parts": \[ { "text": "\* User asks: \\"Who are you\\"\\n \* ", "thought": **true** } \], "role": "model" }, "index": 0 } \], "usageMetadata": { "promptTokenCount": 10, "totalTokenCount": 24, "promptTokensDetails": \[ { "modality": "TEXT", "tokenCount": 10 } \], "thoughtsTokenCount": 14 }, "modelVersion": "gemma-4-31b-it", "responseId": "gKflaeq2J6Pw4-EP9--piAU" }, { "candidates": \[ { "content": { "parts": \[ { "text": "Constraint: Generate only a valid, parseable JSON.\\n \* Constraint: No comments inside JSON.\\n ", "thought": **true** } \], "role": "model" }, "index": 0 } \], "usageMetadata": { "promptTokenCount": 10, "totalTokenCount": 48, "promptTokensDetails": \[ { "modality": "TEXT", "tokenCount": 10 } \], "thoughtsTokenCount": 38 }, "modelVersion": "gemma-4-31b-it", "responseId": "gKflaeq2J6Pw4-EP9--piAU" }, { "candidates": \[ { "content": { "parts": \[ { "text": "\* Constraint: No extra text outside JSON.\\n \* Constraint: If multiple, return a single array.", "thought": **true** } \], "role": "model" }, "index": 0 } \], "usageMetadata": { "promptTokenCount": 10, "totalTokenCount": 72, "promptTokensDetails": \[ { "modality": "TEXT", "tokenCount": 10 } \], "thoughtsTokenCount": 62 }, "modelVersion": "gemma-4-31b-it", "responseId": "gKflaeq2J6Pw4-EP9--piAU" }, { "candidates": \[ { "content": { "parts": \[ { "text": "\\n\\n \* Identity: AI assistant.\\n \* Role: Helpful, informative, and versatile.\\n\\n ", "thought": **true** } \], "role": "model" }, "index": 0 } \], "usageMetadata": { "promptTokenCount": 10, "totalTokenCount": 96, "promptTokensDetails": \[ { "modality": "TEXT", "tokenCount": 10 } \], "thoughtsTokenCount": 86 }, "modelVersion": "gemma-4-31b-it", "responseId": "gKflaeq2J6Pw4-EP9--piAU" }, { "candidates": \[ { "content": { "parts": \[ { "text": "\* Option 1: \`{\\"answer\\": \\"I am an AI assistant.\\"}\`\\n \* Option ", "thought": **true** } \], "role": "model" }, "index": 0 } \], "usageMetadata": { "promptTokenCount": 10, "totalTokenCount": 120, "promptTokensDetails": \[ { "modality": "TEXT", "tokenCount": 10 } \], "thoughtsTokenCount": 110 }, "modelVersion": "gemma-4-31b-it", "responseId": "gKflaeq2J6Pw4-EP9--piAU" }, { "candidates": \[ { "content": { "parts": \[ { "text": "2: \`{\\"identity\\": \\"AI Assistant\\", \\"description\\": \\"A large language model trained by Google.\\"}\`\\n\\n", "thought": **true** } \], "role": "model" }, "index": 0 } \], "usageMetadata": { "promptTokenCount": 10, "totalTokenCount": 144, "promptTokensDetails": \[ { "modality": "TEXT", "tokenCount": 10 } \], "thoughtsTokenCount": 134 }, "modelVersion": "gemma-4-31b-it", "responseId": "gKflaeq2J6Pw4-EP9--piAU" }, { "candidates": \[ { "content": { "parts": \[ { "text": " \* Valid JSON? Yes.\\n \* Double-quoted strings? Yes.\\n \* No", "thought": **true** } \], "role": "model" }, "index": 0 } \], "usageMetadata": { "promptTokenCount": 10, "totalTokenCount": 168, "promptTokensDetails": \[ { "modality": "TEXT", "tokenCount": 10 } \], "thoughtsTokenCount": 158 }, "modelVersion": "gemma-4-31b-it", "responseId": "gKflaeq2J6Pw4-EP9--piAU" }, { "candidates": \[ { "content": { "parts": \[ { "text": " comments? Yes.\\n \* No extra text? Yes.", "thought": **true** } \], "role": "model" }, "index": 0 } \], "usageMetadata": { "promptTokenCount": 10, "totalTokenCount": 182, "promptTokensDetails": \[ { "modality": "TEXT", "tokenCount": 10 } \], "thoughtsTokenCount": 172 }, "modelVersion": "gemma-4-31b-it", "responseId": "gKflaeq2J6Pw4-EP9--piAU" }, { "candidates": \[ { "content": { "parts": \[ { "text": "{\\n \\"identity\\": \\"AI assistant" } \], "role": "model" }, "index": 0 } \], "usageMetadata": { "promptTokenCount": 10, "candidatesTokenCount": 9, "totalTokenCount": 191, "promptTokensDetails": \[ { "modality": "TEXT", "tokenCount": 10 } \], "thoughtsTokenCount": 172 }, "modelVersion": "gemma-4-31b-it", "responseId": "gKflaeq2J6Pw4-EP9--piAU" }, { "candidates": \[ { "content": { "parts": \[ { "text": "\\",\\n \\"description\\": \\"A large language model trained by Google.\\"\\n}" } \], "role": "model" }, "index": 0 } \], "usageMetadata": { "promptTokenCount": 10, "candidatesTokenCount": 26, "totalTokenCount": 208, "promptTokensDetails": \[ { "modality": "TEXT", "tokenCount": 10 } \], "thoughtsTokenCount": 172 }, "modelVersion": "gemma-4-31b-it", "responseId": "gKflaeq2J6Pw4-EP9--piAU" }, { "candidates": \[ { "content": { "parts": \[ { "text": "" } \], "role": "model" }, "finishReason": "STOP", "index": 0 } \], "usageMetadata": { "promptTokenCount": 10, "candidatesTokenCount": 26, "totalTokenCount": 208, "promptTokensDetails": \[ { "modality": "TEXT", "tokenCount": 10 } \], "thoughtsTokenCount": 172 }, "modelVersion": "gemma-4-31b-it", "responseId": "gKflaeq2J6Pw4-EP9--piAU" } ``` ] as you can see thinking is enabled. The doco says it is disabled by default. Anyone know how to disabled it? --- _[View the full topic](https://discuss.ai.google.dev/t/cant-turn-off-thinking-mode-using-gemma4/141243)._