Creator
DeepSeek
Released
2025-05-28
Intelligence
20.1
Artificial Analysis Index
Coding
—
Artificial Analysis Index
In $/1M
$1.35
input tokens
Out $/1M
$4.20
output tokens
Blended $/1M
$2.06
3:1 blended
Speed
0
tokens / sec
Profile
License, openness, and modality — the governance layer.
License
MIT
commercial OK
Weights
Open
downloadable
Modalities
Text
Parameters
671B
total
DeepSeek-R1 is a reasoning-focused language model from DeepSeek that features advanced thinking capabilities. It serves as the foundation for DeepSeek's reasoning model family and pioneered their thinking mode approach for complex problem-solving tasks.
Benchmark breakdown
Independent evaluation scores, normalized to 0–100.
GPQA Diamond
81
Humanity’s Last Exam
15
MMLU-Pro
85
SciCode
40
LiveCodeBench
77
MATH-500
98
AIME 2025
76
τ²-Bench (agentic)
37
Terminal-Bench Hard
16
IFBench
40