Creator
DeepSeek
Released
2025-01-20
Intelligence
6.4
Artificial Analysis Index
Coding
—
Artificial Analysis Index
In $/1M
$0.00
input tokens
Out $/1M
$0.00
output tokens
Blended $/1M
$0.00
3:1 blended
Speed
0
tokens / sec
Profile
License, openness, and modality — the governance layer.
License
MIT
commercial OK
Weights
Open
downloadable
Modalities
—
Parameters
8B
total
DeepSeek-R1 is the first-generation reasoning model built atop DeepSeek-V3 (671B total parameters, 37B activated per token). It incorporates large-scale reinforcement learning (RL) to enhance its chain-of-thought and reasoning capabilities, delivering strong performance in math, code, and multi-step reasoning tasks.
Capability profile
Category strength across 20 domains, via LLM Stats.
math
80general
60reasoning
60biology
50
physics
50chemistry
50code
40
Benchmark breakdown
Independent evaluation scores, normalized to 0–100.
GPQA Diamond
30
Humanity’s Last Exam
4
MMLU-Pro
54
SciCode
12
LiveCodeBench
23
MATH-500
85
AIME 2025
41
τ²-Bench (agentic)—
Terminal-Bench Hard—
IFBench
18