Claude Sonnet 5 scores 44.2 on the Respan Index (#35), ahead of DeepSeek V4 Flash at 43.3 (#37). Claude Sonnet 5 leads on 14 of the 21 benchmarks both report.
* Estimated: no published scores in that category. How the Respan Index works
| GPQA Diamond |
| 90.5% |
| 91% |
| LiveBench Agentic Coding (LiveBench) | 59.4% | 46.8% |
| LiveBench Coding (LiveBench) | 80.7% | 75% |
| LiveBench Data Analysis (LiveBench) | 71.7% | 79.3% |
| LiveBench Instruction Following (LiveBench) | 63.9% | 65.5% |
| LiveBench Language (LiveBench) | 75% | 79.2% |
| LiveBench Mathematics (LiveBench) | 92.9% | 86.8% |
| LiveBench Reasoning (LiveBench) | 88.7% | 86.6% |
| Mystery Game Puzzles (Epoch AI) | 35% | 34% |
| OTIS Mock AIME 2024-2025 (Epoch AI) | 94.7% | 94.4% |
| SimpleQA Verified | 33.7% | 33.6% |
| Terminal-Bench 4.0 (Vals AI) | 9.6% | 18.7% |
| WeirdML (Håvard Tveit Ihle) | 68.8% | 63% |
| LMArena Elo | 1462 | 1436 |
| LMArena WebDev (LMArena) | 1539 | 1581 |
| SimpleBench (SimpleBench) | 60.6% | 61.1% |
| Toolathlon-Verified (HKUST) | 71.6% | 70.7% |
Higher is better. Each score links to where it was published.