GPT-6 Astra scores 70.5 on the Respan Index (#2), ahead of Claude Opus 4.8 at 52.2 (#16). GPT-6 Astra leads on 25 of the 28 benchmarks both report. Claude Opus 4.8 is 2.0x cheaper per token at list price. GPT-6 Astra has the larger context window (1.1M tokens).
* Estimated: no published scores in that category. How the Respan Index works
| 56.1% |
| 97.6% |
| FrontierMath Tiers 1-3 v2 (Epoch AI) | 80% | 93.7% |
| Furniture Assembly (Epoch AI) | 42.5% | 80% |
| GPQA Diamond | 93.6% | 96% |
| LiveBench Agentic Coding (LiveBench) | 50.5% | 57.3% |
| LiveBench Coding (LiveBench) | 81.8% | 80.4% |
| LiveBench Data Analysis (LiveBench) | 66% | 83% |
| LiveBench Instruction Following (LiveBench) | 72% | 75.6% |
| LiveBench Language (LiveBench) | 79.7% | 89.4% |
| LiveBench Mathematics (LiveBench) | 94.3% | 96.8% |
| LiveBench Reasoning (LiveBench) | 89.2% | 92.7% |
| Mystery Game Puzzles (Epoch AI) | 36% | 84% |
| OTIS Mock AIME 2024-2025 (Epoch AI) | 98.3% | 100% |
| SAGE (Vals AI) | 54.8% | 46.4% |
| SimpleQA Verified | 53% | 75.6% |
| Terminal-Bench 4.0 (Vals AI) | 23.2% | 59.6% |
| Vending-Bench 2 (Andon Labs) | 5787.43 | 15514.7 |
| WeirdML (Håvard Tveit Ihle) | 82.9% | 93.6% |
| LMArena Agent (LMArena) | 0.0664 | 0.1227 |
| LMArena Elo | 1453 | 1477 |
| LMArena Vision (LMArena) | 1286 | 1284 |
| LMArena WebDev (LMArena) | 1556 | 1788 |
| GSO Opt@1 (GSO) | 47.1% | 77.5% |
| SimpleBench (SimpleBench) | 64.8% | 86.5% |
Higher is better. Each score links to where it was published.