Claude Fable 5.1 scores 69.4 on the Respan Index (#4), ahead of Qwen3.8-Max at 50.3 (#22). Claude Fable 5.1 leads on 19 of the 22 benchmarks both report. Qwen3.8-Max is 6.7x cheaper per token at list price.
* Estimated: no published scores in that category. How the Respan Index works
| FrontierSWE V2 (Proximal Labs) |
| 56.3% |
| 17.8% |
| Furniture Assembly (Epoch AI) | 70% | 20% |
| LiveBench Agentic Coding (LiveBench) | 66.1% | 64.6% |
| LiveBench Coding (LiveBench) | 86.4% | 72.9% |
| LiveBench Data Analysis (LiveBench) | 80.3% | 78.4% |
| LiveBench Instruction Following (LiveBench) | 73% | 74.1% |
| LiveBench Language (LiveBench) | 89.5% | 79.7% |
| LiveBench Mathematics (LiveBench) | 97% | 91.3% |
| LiveBench Reasoning (LiveBench) | 91.7% | 88.2% |
| Mystery Game Puzzles (Epoch AI) | 58% | 38% |
| OTIS Mock AIME 2024-2025 (Epoch AI) | 100% | 99.4% |
| SAGE (Vals AI) | 48.5% | 51.3% |
| SimpleQA Verified | 70.8% | 45.8% |
| Terminal-Bench 4.0 (Vals AI) | 58.1% | 34.3% |
| LMArena Agent (LMArena) | 0.1431 | 0.0251 |
| LMArena Elo | 1501 | 1481 |
| LMArena Vision (LMArena) | 1288 | 1301 |
| LMArena WebDev (LMArena) | 1749 | 1671 |
Higher is better. Each score links to where it was published.