DeepSeek V3 (Dec '24): benchmark results
Provider: DeepSeek. Released 2024-12-26. Access: Open.
Unified ELO 1557 ± 1, rank #415 of 1392 rated models, from 20 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AA Omniscience - Humanities & Social Sciences | 27.1 | Accuracy (%) | 69.2 |
| AA Omniscience - Law | 18.4 | Accuracy (%) | 68.7 |
| AA Omniscience - Health | 25.5 | Accuracy (%) | 68.2 |
| AA Omniscience - Business | 21.5 | Accuracy (%) | 67 |
| AA-Omniscience Accuracy | 25.45 | Accuracy (%) | 64.9 |
| AA Omniscience - Software Engineering (SWE) | 33.5 | Accuracy (%) | 64.1 |
| AA Omniscience | -41.65 | Score | 46.2 |
| AA Omniscience - Science, Engineering & Mathematics | 26.7 | Accuracy (%) | 45.6 |
| Artificial Analysis Intelligence Index | 8.27 | Intelligence Index | 43.2 |
| AA Terminal-Bench Hard | 6.82 | Accuracy (%) | 37.1 |
| AA Long Context Reasoning | 29.33 | Accuracy (%) | 31.7 |
| AA GPQA Diamond | 55.66 | Accuracy (%) | 29.4 |
Interactive version: theaggregate.ai/model?slug=deepseek-v3-dec-24 · How It Works · Data refreshed daily, snapshot 2026-09-05.