Claude Sonnet 5 scores 44.2 on the Respan Index (#35), ahead of GPT-5.2 at 43.5 (#36). Claude Sonnet 5 leads on 12 of the 23 benchmarks both report. Claude Sonnet 5 is 1.2x cheaper per token at list price. Claude Sonnet 5 has the larger context window (1M tokens).
* Estimated: no published scores in that category. How the Respan Index works
| 59.4% |
| 50.3% |
| LiveBench Coding (LiveBench) | 80.7% | 76.1% |
| LiveBench Data Analysis (LiveBench) | 71.7% | 78.2% |
| LiveBench Instruction Following (LiveBench) | 63.9% | 61.8% |
| LiveBench Language (LiveBench) | 75% | 79.8% |
| LiveBench Mathematics (LiveBench) | 92.9% | 93.2% |
| LiveBench Reasoning (LiveBench) | 88.7% | 83.2% |
| Mystery Game Puzzles (Epoch AI) | 35% | 23% |
| OTIS Mock AIME 2024-2025 (Epoch AI) | 94.7% | 96.1% |
| SAGE (Vals AI) | 48.9% | 49.3% |
| SimpleQA Verified | 33.7% | 37.1% |
| Vending-Bench 2 (Andon Labs) | 6377.7 | 3591.33 |
| WeirdML (Håvard Tveit Ihle) | 68.8% | 72.2% |
| LMArena Elo | 1462 | 1412 |
| LMArena Vision (LMArena) | 1265 | 1243 |
| LMArena WebDev (LMArena) | 1539 | 1416 |
| GSO Opt@1 (GSO) | 36.3% | 26.5% |
| SimpleBench (SimpleBench) | 60.6% | 45.8% |
| SWE-bench Verified | 85.2% | 80% |
Higher is better. Each score links to where it was published.