Claude Opus 5 scores 63.9 on the Respan Index (#7), ahead of Grok 4.6 at 51.0 (#19). Claude Opus 5 leads on 25 of the 28 benchmarks both report. Grok 4.6 is 3.3x cheaper per token at list price. Claude Opus 5 has the larger context window (1M tokens).
* Estimated: no published scores in that category. How the Respan Index works
| 73.2% |
| 31.7% |
| FrontierMath Tiers 1-3 v2 (Epoch AI) | 85.6% | 66% |
| FrontierSWE V2 (Proximal Labs) | 52% | 25.3% |
| Furniture Assembly (Epoch AI) | 60.8% | 40% |
| GPQA Diamond | 93.9% | 94% |
| LiveBench Agentic Coding (LiveBench) | 65.2% | 57% |
| LiveBench Coding (LiveBench) | 81.4% | 76.8% |
| LiveBench Data Analysis (LiveBench) | 74.6% | 73.9% |
| LiveBench Instruction Following (LiveBench) | 63.8% | 71.9% |
| LiveBench Language (LiveBench) | 88.7% | 83.7% |
| LiveBench Mathematics (LiveBench) | 95.7% | 92.6% |
| LiveBench Reasoning (LiveBench) | 91.2% | 90.5% |
| Mystery Game Puzzles (Epoch AI) | 59% | 34% |
| OTIS Mock AIME 2024-2025 (Epoch AI) | 98.9% | 99.2% |
| SAGE (Vals AI) | 49.4% | 28.9% |
| SimpleQA Verified | 59.9% | 49.3% |
| Terminal-Bench 4.0 (Vals AI) | 53.5% | 17.2% |
| Vending-Bench 2 (Andon Labs) | 11181.87 | 9047.03 |
| WeirdML (Håvard Tveit Ihle) | 91.8% | 67.3% |
| LMArena Agent (LMArena) | 0.0867 | 0.0128 |
| LMArena Elo | 1490 | 1454 |
| LMArena Vision (LMArena) | 1289 | 1264 |
| LMArena WebDev (LMArena) | 1695 | 1620 |
| SimpleBench (SimpleBench) | 80.6% | 75.9% |
Higher is better. Each score links to where it was published.