DeepSeek V3.1 (Non-reasoning)
Unknown SizeBy DeepSeek • Released 2025-08-21
Capability Radar
Avg Score
42
Across all benchmarks
Participated
12
Benchmarks
Benchmark Performance
| Benchmark | Category | Score |
|---|---|---|
| MMLU-Pro | Knowledge | 83.3 |
| GPQA Diamond | Knowledge | 73.5 |
| LiveCodeBench | Coding | 57.7 |
| AIME 2025 | Reasoning | 49.7 |
| LCR | Long-Context Reasoning | 45 |
| IFBench | Agent | 37.8 |
| SciCode | Reasoning Knowledge | 36.7 |
| 𝜏²-Bench Telecom | Reasoning Knowledge | 34.8 |
| Artificial Analysis Coding Index | Coding | 28.4 |
| Artificial Analysis Intelligence Index | Knowledge | 28 |
| Terminal-Bench Hard | Agent Coding | 24.2 |
| HLE | Knowledge Multi-Modal | 6.3 |