DeepSeekDeepSeek
DeepSeek V3 0324
DeepSeek · DeepSeek. Released Mar 24, 2025.
DeepSeek V3 0324 is a model from DeepSeek in the DeepSeek family, released Mar 24, 2025. evals.report tracks 15 reported DeepSeek V3 0324 benchmark scores across LiveCodeBench Pro, LMArena, WeirdML, Epoch Capabilities Index, Aider Polyglot, MMLU-Pro, GDPval, LiveCodeBench, and 7 more — each shown with its benchmark, metric, source status, and date, and never combined into a single ranking.
Open15 results
Benchmark results 15
Compare this model| Benchmark | Category | Score | Metric | Status | Date | |
|---|---|---|---|---|---|---|
| LiveCodeBench Pro | Coding | 1124 | Codeforces Elo | Official | Mar 24, 2025 | Details |
| LMArena | Chat preference | 1374 | source-defined rating | Official | Mar 24, 2025 | Details |
| WeirdML | Coding | 36.1% | average accuracy | Official | Mar 24, 2025 | Details |
| Epoch Capabilities Index | Reasoning | 137.2 | Index | Official | Mar 24, 2025 | Details |
| Aider Polyglot | Coding | 55.1% | % correct | Official | Mar 24, 2025 | Details |
| MMLU-Pro | Reasoning | 81.9% | accuracy | Verified | Mar 24, 2025 | Details |
| GDPval | Agents | 407 | Elo | Official | Mar 24, 2025 | Details |
| LiveCodeBench | Coding | 40.5% | Pass@1 | Unverified | Mar 24, 2025 | Details |
| SciCode | Coding | 35.8% | accuracy | Unverified | Mar 24, 2025 | Details |
| EQ-Bench Creative Writing v3 | Chat preference | 1474 | Elo | Verified | Mar 24, 2025 | Details |
| Design Arena | Chat preference | 1163 | Elo | Verified | Mar 24, 2025 | Details |
| MASK (Model Alignment between Statements and Knowledge) | Other | 44.53 | Honesty score | Verified | Mar 24, 2025 | Details |
| BigCodeBench | Coding | 35.8% | calibrated Pass@1 | Verified | Mar 24, 2025 | Details |
| PutnamBench | Reasoning | 0 | Problems solved | Verified | Mar 24, 2025 | Details |
| PolyMath: Evaluating Mathematical Reasoning in Multilingual Contexts | Reasoning | 30.7 | Difficulty-Weighted Accuracy (DW-ACC) | Verified | Mar 24, 2025 | Details |