DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning (arXiv 2501.12948) Jan 20, 2025
Capabilities
- AIME 2024 28.9
- MATH-500 83.9
- GPQA Diamond 33.8
- LiveCodeBench 16.9
- Codeforces rating 954
DeepSeek-R1-Distill-Qwen-1.5B
DeepSeek-R1-Distill-Qwen-1.5B.
17
Benchmarks with a score
0
Ranked first, out of five or more models
0
In the top three
DeepSeek Jan 2025 DeepSeek-R1-Distill-Qwen-1.5B
Open weights
No Epoch Capabilities Index score yet
- GPQA Diamond 33.6% #201 /229
- BaxBench 0% #34 /34
- MathIF 17.1% #12 /12
Where it ranks
Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.
Knowledge and language
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Seal-0 | 2.70% | #5 of 11 | Official leaderboard | |
| MuSR | 36.4% | #81 of 98 | Official leaderboard | |
| Global PIQA | 27.9% | #43 of 44 | Official leaderboard | |
| MathIF | 17.1% | #12 of 12 | Official leaderboard | |
| MT-AIME2024 | 17.2% | #4 of 4 | Official leaderboard |
Math and science
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| OTIS Mock AIME 2024-2025 | 21.4% | #153 of 212 | Epoch AI run | |
| MATH-500 | 83.9% | #56 of 70 | Lab model card | |
| GPQA Diamond | 33.6% | #201 of 229 | Epoch AI run | |
| AIME 2024 | 28.9% | #81 of 88 | Lab model card | |
| HMMT 2025 | 11.7% | #50 of 53 | Official leaderboard | |
| AIME 2025 | 20% | #51 of 53 | Official leaderboard |
Coding
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Codeforces rating | 954 | #44 of 47 | Lab model card | |
| LiveCodeBench | 16.9% | #51 of 54 | Lab model card | |
| BigCodeBench | 7.50% | #69 of 69 | Official leaderboard |
Professional work
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| BRIDGE | 14.9 | #39 of 39 | Official leaderboard |
Safety and security
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| BaxBench | 0% | #34 of 34 | Official leaderboard |
Other
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Open LLM Leaderboard v2 | 10.3 | #86 of 99 | Official leaderboard |
Its lab's number next to a leaderboard
Where the benchmark's own leaderboard, or Epoch AI's run, also has this model. Gaps mostly come from different harnesses, effort levels and task subsets. Every pair.
| Benchmark | Lab says | Board says | Gap |
|---|---|---|---|
| GPQA Diamond | 33.8% GPQA Diamond pass@1, DeepSeek-R1-Distill-Qwen-1.5B (Table 15) | 33.6% Epoch AI run | +0.2 pts |
Its system card
Every result in the longer documents the lab published with it. 5 of them aren't in the launch post.