DeepSeek-V3 is the stronger all-rounder.
Scores updated · 27 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.
Reasoning rests on a single test.
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-V3 pulls ahead
- Recent programming contest problemsLiveCodeBench v6+20.6points ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+17.4points ahead
- Fixes real GitHub issues in Python projectsSWE-bench Verified+13.8points ahead
Where Kimi K2 Base pulls ahead
No clear win on a test scored out of 100.
Every test, side by side
All 27 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingDeepSeek-V3
- LiveCodeBench v6DeepSeek-V3 by 20.646.926.3+20.6
- SWE-bench VerifiedDeepSeek-V3 by 13.84228.2+13.8
Other results24 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.
- OpenBookQADeepSeek-V3 by 44.695.450.8+44.6
- MATHDeepSeek-V3 by 2090.270.2+20
- MBPPKimi K2 Base by 13.675.489+13.6
- HumanEvalKimi K2 Base by 1365.278.2+13
- SimpleQAKimi K2 Base by 10.424.935.3+10.4
- SuperGPQADeepSeek-V3 by 953.744.7+9
- DROPDeepSeek-V3 by 891.683.6+8
- MMLU-ProDeepSeek-V3 by 6.775.969.2+6.7
- Chinese SimpleQA (C-SimpleQA)Kimi K2 Base by 6.66874.6+6.6
- C-EvalKimi K2 Base by 686.592.5+6
- HellaSwagKimi K2 Base by 5.788.994.6+5.7
- MGSMKimi K2 Base by 5.479.885.2+5.4
- MBPP+ (EvalPlus-augmented)DeepSeek-V3 by 578.873.8+5
- CMMLUKimi K2 Base by 2.188.890.9+2.1
- GSM8KDeepSeek-V3 by 1.99492.1+1.9
- MMMLUDeepSeek-V3 by 1.879.477.6+1.8
- BBHKimi K2 Base by 1.287.588.7+1.2
- MMLU-ReduxKimi K2 Base by 1.189.190.2+1.1
- ARC-Challengetie95.396.2tie
- CRUXEval-I (input prediction)tie67.368tie
- MMLUtie88.587.8tie
- WinoGrandetie84.985.3tie
- CRUXEval-O (output prediction)tie69.869.6tie
- PIQAtie84.784.9tie
Questions people ask
Which is better, DeepSeek-V3 or Kimi K2 Base?
DeepSeek-V3 wins all two areas where both have results: coding and reasoning. Kimi K2 Base wins none.
Which is better for coding?
DeepSeek-V3. It wins 2 of the 2 coding tests both models report; Kimi K2 Base wins none.
How do you compare the two?
We use the 27 benchmark tests both models have published scores on. The verdict counts the 3 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 24 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.