DE

DeepSeekMath-V2

Reasoning modelDeepSeekMath V2

DeepSeekMath-V2

Release date: 2025-11-27Updated: 2026-06-15 07:18:20.831731
Parameters
685B
Context length
108K
Chinese support
Supported
Reasoning ability

DeepSeekMath-V2 is an AI model published by DeepSeek-AI, released on 2025-11-27, for Reasoning model, with 685B parameters, and 108K context length, under the Apache 2.0 license, with a 99.00 score on IMO-ProofBench.

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

DeepSeekMath-V2

Model basics

Reasoning traces
Supported
Thinking modes
Thinking Level · Extended (Default)
Context length
108K tokens
Max output length
No data
Model type
Reasoning model
Modality (in / out)
Text → Text
Release date
2025-11-27
Model file size
No data
MoE architecture
Yes
Total params / Active params
685B / 37B
Knowledge cutoff
No data
DeepSeekMath-V2

Open source & experience

Code license
Weights license
Apache 2.0- 免费商用授权
Live demo
No live demo
DeepSeekMath-V2

Official resources

DeepSeekMath-V2

API details

API speed
3/5
No public API pricing yet.
DeepSeekMath-V2

Benchmark Results

DeepSeekMath-V2 currently shows benchmark results led by IMO-ProofBench (1 / 16, score 99), IMO-ProofBench Advanced (1 / 8, score 61.90). This page also consolidates core specs, context limits, and API pricing so you can evaluate the model from benchmark results and deployment constraints together.

Thinking

Math and Reasoning

2 evaluations
Benchmark / mode
Score
Rank/total

Compare with other models

No curated comparisons for this model yet.

Want a custom combination? Open the compare tool

DeepSeekMath-V2

Publisher

DeepSeekMath-V2

Model Overview

DeepSeekMath-V2 is an AI model published by DeepSeek-AI, released on 2025-11-27, for Reasoning model, with 685B parameters, and 108K context length, under the Apache 2.0 license, with a 99.00 score on IMO-ProofBench.

DataLearner on WeChat

Follow DataLearner on WeChat for AI model updates and research notes.

DataLearner WeChat QR code