Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology

Gemini 2.5 Deep Think

Model basics

Reasoning traces

Supported

Context length

1000K tokens

Max output length

16384 tokens

Model type

聊天大模型

Release date

2025-08-01

Model file size

No data

MoE architecture

Total params / Active params

0.0B / N/A

Knowledge cutoff

No data

Inference modes

深度思考（Deeper Thinking Mode）

Gemini 2.5 Deep Think

Open source & experience

Code license

不开源

Weights license

不开源- 不开源

GitHub repo

https://github.com/QwenLM/qwen-code

Hugging Face

https://huggingface.co/Qwen/Qwen3-Coder-30B-A3B-Instruct

Live demo

https://chat.qwen.ai/

Gemini 2.5 Deep Think

Official resources

Paper

Try Deep Think in the Gemini app

DataLearnerAI blog

No blog post yet

Gemini 2.5 Deep Think

API details

API speed

3/5

💡Default unit: $/1M tokens. If vendors use other units, follow their published pricing.

Standard pricingStandard

Modality	Input	Output
Text	$0.4	$1.6

Gemini 2.5 Deep Think

Benchmark Results

Thinking

综合评估

1 evaluations

Benchmark / mode

Score

Rank/total

HLEDeep thinking

34.80

28 / 105

编程与软件工程

1 evaluations

Benchmark / mode

Score

Rank/total

LiveCodeBenchDeep thinking

87.60

2 / 103

数学推理

4 evaluations

Benchmark / mode

Score

Rank/total

AIME2025Deep thinking

99.20

10 / 105

IMO-ProofBenchThinking

37.60

5 / 16

FrontierMathDeep thinking

4 / 52

FrontierMath - Tier 4Normal

10.40

9 / 32

查看评测深度分析与其他模型对比

Gemini 2.5 Deep Think

Publisher

Google Deep Mind

View publisher details

Gemini 2.5 Deep Think

Model Overview

Gemini 2.5 Deep Think并不是Google发布的全新模型，而是Geimini 2.5 Pro的一个并行思考推理模式。模型来自一个获得数学奥赛金牌模型的变种版本，获取奥赛金牌的那个模型需要几个小时完成复杂的数学任务，但是本模型更快，使用并行思考的模型，针对同一个问题一次性生成多个答案之后然后对这些答案总结。

DataLearner 官方微信

欢迎关注 DataLearner 官方微信，获得最新 AI 技术推送