DeepSeek V3 (Dec '24): benchmark results
Provider: DeepSeek. Released 2024-12-26. Access: Open.
Aggregate Rating 1559 ± 42, rank #587 of 1656 rated models, from 69 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AA-Omniscience Accuracy | 25.45 | Accuracy (%) | 59.4 |
| AA-Omniscience Index - Software Engineering (SWE) - Kotlin | -24 | Omniscience Index | 56.4 |
| AA-Omniscience Index - Software Engineering (SWE) - Dart | -28 | Omniscience Index | 54.7 |
| AA-Omniscience Index - Software Engineering (SWE) - Java | -40 | Omniscience Index | 54.5 |
| AA-Omniscience Index - Software Engineering (SWE) - TypeScript | -30 | Omniscience Index | 50.3 |
| AA MMLU-Pro | 75.22 | Accuracy (%) | 50 |
| AA-Omniscience Index - Software Engineering (SWE) - R | -48 | Omniscience Index | 47.3 |
| AA-Omniscience Index - Software Engineering (SWE) - Python | -36 | Omniscience Index | 46.7 |
| AA-Omniscience Index - Software Engineering (SWE) | -31.5 | Omniscience Index | 46.2 |
| AA-Omniscience Index - Software Engineering (SWE) - Go | -38 | Omniscience Index | 46.1 |
| AA-Omniscience Index - Software Engineering (SWE) - C | -14 | Omniscience Index | 45.6 |
| AA-Omniscience Index - Software Engineering (SWE) - Swift | -16 | Omniscience Index | 45 |
Interactive version: theaggregate.ai/model?slug=deepseek-v3-dec-24 · How It Works · Data refreshed daily, snapshot 2026-10-20.