DeepSeek-R1 is the stronger all-rounder.
Scores updated · 34 tests both models report · How we compare
Where each one wins
Tests won in each of the two areas where both have results. Each piece is one test, so a longer bar means more evidence; grey means the two scored within a point of each other.
What it costs
Prices per million tokens, roughly 750,000 words. The bars show the cost of a million tokens read plus a million written.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where DeepSeek-R1 pulls ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+10.5points ahead
- Fixes real GitHub issues in Python projectsSWE-bench Verified+7.6points ahead
- Code for real scientific research problemsSciCode+6points ahead
Where o1-mini pulls ahead
No clear win on a test scored out of 100.
Every test, side by side
All 34 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingDeepSeek-R1
- SWE-bench VerifiedDeepSeek-R1 by 7.649.241.6+7.6
- SciCodeDeepSeek-R1 by 638.332.3+6
ReasoningDeepSeek-R1
- GPQA DiamondDeepSeek-R1 by 10.570.860.3+10.5
- Humanity's Last ExamDeepSeek-R1 by 4.98.53.6+4.9
Other results30 tests, not counted
Tests outside the eight areas. They are not counted above: several are summary scores built from other tests, or the same test under another name.
- HarmBencho1-mini by 40.647.988.5+40.6
- AlpacaEval2.0 (LC-winrate)DeepSeek-R1 by 29.887.657.8+29.8
- Chinese SimpleQA (C-SimpleQA)DeepSeek-R1 by 23.463.740.3+23.4
- SimpleQADeepSeek-R1 by 23.130.17+23.1
- C-EvalDeepSeek-R1 by 22.991.868.9+22.9
- CodeforcesDeepSeek-R1 by 209 rating points20291820+209 rating
- Codeforces (Rating)DeepSeek-R1 by 209 rating points20291820+209 rating
- Aider-PolyglotDeepSeek-R1 by 20.453.332.9+20.4
- AIME 2025DeepSeek-R1 by 19.37050.7+19.3
- AIME 2024DeepSeek-R1 by 16.279.863.6+16.2
- LiveCodeBench (Pass@1-COT)DeepSeek-R1 by 12.165.953.8+12.1
- CNMO 2024DeepSeek-R1 by 11.278.867.6+11.2
- DROP (3-shot F1)DeepSeek-R1 by 8.392.283.9+8.3
- AIR-Bench 2024DeepSeek-R1 by 7.652.945.3+7.6
- MATHDeepSeek-R1 by 7.397.390+7.3
- MATH-500 (EM)DeepSeek-R1 by 7.397.390+7.3
- MMLU-ReduxDeepSeek-R1 by 6.292.986.7+6.2
- FRAMES (Acc.)DeepSeek-R1 by 5.682.576.9+5.6
- MMLUDeepSeek-R1 by 5.690.885.2+5.6
- MMLU-ProDeepSeek-R1 by 3.78480.3+3.7
- CLUEWSCDeepSeek-R1 by 2.992.889.9+2.9
- Codeforces (Percentile)DeepSeek-R1 by 2.996.393.4+2.9
- XSTesto1-mini by 2.694.497+2.6
- ARC-AGI-1DeepSeek-R1 by 1.815.814+1.8
- AA IntelligenceDeepSeek-R1 by 1.611.49.8+1.6
- IFEvalo1-mini by 1.583.384.8+1.5
- anthropic_red_teamo1-mini by 1.197.298.3+1.1
- simple_safety_testsDeepSeek-R1 by 19897+1
- Arena-Hard (GPT-4-1106 judge)tie92.392tie
- bbqtie96.696.9tie
Questions people ask
Which is better, DeepSeek-R1 or o1-mini?
DeepSeek-R1 wins all two areas where both have results: coding and reasoning. o1-mini wins none.
Which is better for coding?
DeepSeek-R1. It wins 2 of the 2 coding tests both models report; o1-mini wins none.
How do you compare the two?
We use the 34 benchmark tests both models have published scores on. The verdict counts the 4 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 30 are listed but not counted, because several are summary scores or repeat a test. Each score is the one shown on the model's own page.