GPT-6 Astra scores 70.5 on the Respan Index (#2), ahead of Kimi K3 at 50.6 (#21). GPT-6 Astra leads on 22 of the 26 benchmarks both report. Kimi K3 is 3.3x 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
| 93.7% |
| 72.2% |
| FrontierSWE V2 (Proximal Labs) | 65.5% | 25.9% |
| Furniture Assembly (Epoch AI) | 80% | 34.2% |
| GPQA Diamond | 96% | 93.5% |
| LiveBench Agentic Coding (LiveBench) | 57.3% | 62.2% |
| LiveBench Coding (LiveBench) | 80.4% | 81.4% |
| LiveBench Data Analysis (LiveBench) | 83% | 78.7% |
| LiveBench Instruction Following (LiveBench) | 75.6% | 71.4% |
| LiveBench Language (LiveBench) | 89.4% | 85.5% |
| LiveBench Mathematics (LiveBench) | 96.8% | 84.4% |
| LiveBench Reasoning (LiveBench) | 92.7% | 90.7% |
| Mystery Game Puzzles (Epoch AI) | 84% | 26% |
| OTIS Mock AIME 2024-2025 (Epoch AI) | 100% | 97.2% |
| SAGE (Vals AI) | 46.4% | 52.8% |
| SimpleQA Verified | 75.6% | 50.6% |
| Terminal-Bench 4.0 (Vals AI) | 59.6% | 17.2% |
| Vending-Bench 2 (Andon Labs) | 15514.7 | 5165.04 |
| WeirdML (Håvard Tveit Ihle) | 93.6% | 82.6% |
| LMArena Agent (LMArena) | 0.1227 | 0.0418 |
| LMArena Elo | 1477 | 1488 |
| LMArena WebDev (LMArena) | 1788 | 1658 |
| SimpleBench (SimpleBench) | 86.5% | 60.7% |
Higher is better. Each score links to where it was published.