R1
Open weights
DeepSeek's open-weights reasoning model - the January 2025 release that triggered the "DeepSeek moment" in global AI markets.
- Publisher
- DeepSeek
- Family
- Deepseek R
- Params
- 671B total / 37B active
- Context
- 64K
- Input
- Output
- License
- MIT
- Released
- 20 Jan 2025
Cite
Notes
Only stored in your browser.
Attribution
- Benchmark scores
- AiderAAVaultarena-hard-autoLiveBenchOpenRewardAfterQueryVals AIEQ-Bench
Context
64K
Input
$2
/1M tok
Output
$4
/1M tok
Blended
$2.50
#395/ 461/1M · 3:1
Speed
0
#115/ 461tok/s
Price & speed via Artificial Analysis · ranked across tracked models
| Reasoning effort | Index | Price | Speed |
|---|---|---|---|
| may-25 | 20.4 | $1.76 | 0 t/s |
| jan-25 | 18.6 | $2.50 | 0 t/s |
Intelligence Index per reasoning-effort setting, via Artificial Analysis
Reported on 23 evals across 11 domains - best 96.6% on MATH-500 · #19 of 190
Reported eval scores
23MMLU-Pro84.4
LiveBench66.8
IFBench39
MATH-50096.6
LiveCodeBench61.7
AIME202479.8
GPQA Diamond70.8
SciCode35.7
TaxEval v272.3