Claude Opus 4.5 scores 37.5 on the Respan Index (#52), ahead of GPT-5 at 37.2 (#54). Claude Opus 4.5 leads on 11 of the 17 benchmarks both report. GPT-5 is 2.9x cheaper per token at list price. GPT-5 has the larger context window (400K tokens).
* Estimated: no published scores in that category. How the Respan Index works
| 87% |
| 86.2% |
| Mystery Game Puzzles (Epoch AI) | 22% | 23% |
| OTIS Mock AIME 2024-2025 (Epoch AI) | 86.1% | 91.4% |
| SAGE (Vals AI) | 52.1% | 43.7% |
| SimpleQA Verified | 45.7% | 50.1% |
| SWE-bench Verified (Epoch AI) | 76.7% | 73.6% |
| WeirdML (Håvard Tveit Ihle) | 63.7% | 60.7% |
| LMArena Elo | 1474 | 1435 |
| LMArena WebDev (LMArena) | 1493 | 1417 |
| BALROG (BALROG) | 43.5% | 32.8% |
| GSO Opt@1 (GSO) | 24.5% | 5.9% |
| SWE-bench Verified | 80.9% | 74.9% |
| SimpleBench (SimpleBench) | 62% | 56.7% |
Higher is better. Each score links to where it was published.