DeepSeek V3.1

Non-reasoning

Creator

DeepSeek

Released

2025-08-21

Intelligence

28.1

Artificial Analysis Index

Coding

28.4

Artificial Analysis Index

In $/1M

$0.56

input tokens

Out $/1M

$1.67

output tokens

Blended $/1M

$0.83

3:1 blended

Speed

0

tokens / sec

Benchmark breakdown

Independent evaluation scores, normalized to 0–100.

GPQA Diamond
74
Humanity’s Last Exam
6
MMLU-Pro
83
SciCode
37
LiveCodeBench
58
MATH-500
AIME 2025
50
τ²-Bench (agentic)
35
Terminal-Bench Hard
24
IFBench
38
Via Artificial Analysis