DeepSeek-V3.1

Hybrid-reasoning DeepSeek model with thinking and non-thinking modes

Context window
131.1K
tokens in · 8.2K out
Since the last one
150 days
after DeepSeek V3 0324

Since DeepSeek V3 0324

8 releases in the DeepSeek V4 Preview line.

Released150 days later
Context131.1Kwas 163.8K

A newer release is out: DeepSeek V4 Preview.

ECI STANDING#124 of 270
SCORE139.79135.65–143.26 CI
FRONTIER LAGLaunched 8.1 months behindScores about what o1 did in December 2024.
SCORES LIKE THESE MODELS
Kimi K2 (Jul 2025)Moonshot · Jul 2025139.97
Gemini 2.5 Flash (Apr 2025)Google DeepMind · Apr 2025139.84
gpt-oss-120bOpenAI · Aug 2025139.79
Qwen3-30B-A3B-Thinking (Jul 2025)Alibaba · Jul 2025139.55
Full Epoch Capabilities Index ranking ↗

Epoch benchmark boards

Best result on each board it has run. The tick marks the leader.

DTBench
83%#73 of 179
WeirdML
38%#97 of 141
Fiction.LiveBench
53%#40 of 58
SimpleBench
40%#62 of 89

DeepSeek's past year

9 release days in the past year, about one every 41 days. The next is expected around Nov 28, 2026. Forecasts ↗

NovDecJanFebMarAprMayJunJulAugSepOctNov

Specifications

RELEASE DATEAug 21, 2025
WEIGHTSOpen weights
CONTEXT WINDOW131.1K
MAX OUTPUT8.2K
REASONINGNot listed
TOOL CALLINGNot listed
INPUTStext
OUTPUTStext

Models.dev record ↗

Specifications and prices from Models.dev; availability and limits may differ by provider. Benchmarks from Epoch AI (CC BY 4.0), LiveBench and the other sources named above.