Claude Fable 5.1 scores 69.4 on the Respan Index (#4), ahead of Kimi K3 at 50.6 (#21). Claude Fable 5.1 leads on 25 of the 26 benchmarks both report. Kimi K3 is 3.3x cheaper per token at list price. Kimi K3 has the larger context window (1.0M tokens).
* Estimated: no published scores in that category. How the Respan Index works
| FrontierSWE V2 (Proximal Labs) |
| 56.3% |
| 25.9% |
| Furniture Assembly (Epoch AI) | 70% | 34.2% |
| LiveBench Agentic Coding (LiveBench) | 66.1% | 62.2% |
| LiveBench Coding (LiveBench) | 86.4% | 81.4% |
| LiveBench Data Analysis (LiveBench) | 80.3% | 78.7% |
| LiveBench Instruction Following (LiveBench) | 73% | 71.4% |
| LiveBench Language (LiveBench) | 89.5% | 85.5% |
| LiveBench Mathematics (LiveBench) | 97% | 84.4% |
| LiveBench Reasoning (LiveBench) | 91.7% | 90.7% |
| Mystery Game Puzzles (Epoch AI) | 58% | 26% |
| OTIS Mock AIME 2024-2025 (Epoch AI) | 100% | 97.2% |
| SAGE (Vals AI) | 48.5% | 52.8% |
| SimpleQA Verified | 70.8% | 50.6% |
| Terminal-Bench 4.0 (Vals AI) | 58.1% | 17.2% |
| Vending-Bench 2 (Andon Labs) | 5421.56 | 5165.04 |
| WeirdML (Håvard Tveit Ihle) | 92.9% | 82.6% |
| LMArena Agent (LMArena) | 0.1431 | 0.0418 |
| LMArena Elo | 1501 | 1488 |
| LMArena WebDev (LMArena) | 1749 | 1658 |
| SimpleBench (SimpleBench) | 86.6% | 60.7% |
| GDPval-AA v2 | 1853 | 1686 |
| Humanity's Last Exam (no tools) | 60.9% | 43.5% |
Higher is better. Each score links to where it was published.