Across 26 shared benchmarks, Claude Fable 5 scores higher on 25 and Qwen3.5 122B A10B on 1. The widest gap is OmniScience Non-Hallucination, where Claude Fable 5 scores 36.4 against 12.9. Qwen3.5 122B A10B is the cheaper of the two on tracked API pricing ($0.40 against $10.00 per million input tokens).
| Benchmark | Claude Fable 5 | Qwen3.5 122B A10B |
|---|---|---|
| AA Agentic Index | 56.6 | 21.3 |
| AA Intelligence | 62.1 | 32.8 |
| AA-LCR | 76.7 | 70.3 |
| AA-Omniscience | 43.3 | -41.5 |
| arena_vision | 1328 | 1246 |
| Artificial Analysis Coding Index | 76.5 | 45.7 |
| browsecomp | 88 | 63.8 |
| coding_arena_elo | 1626 | 1358 |
| critpt | 28.6 | 0.9 |
| gdpval | 61.5 | 24.3 |
| GPQA Diamond | 92.6 | 86.6 |
| HLE | 64.5 | 47.5 |
| IFBench | 63.5 | 76.1 |
| mathvision | 94.8 | 86.2 |
| MMMU-Pro | 84.2 | 76.9 |
| OmniScience Accuracy | 65.3 | 24.4 |
| OmniScience Non-Hallucination | 36.4 | 12.9 |
| OSWorld-Verified | 85 | 58 |
| scicode | 60.2 | 42 |
| SWE-bench Verified | 95 | 72 |
| TauBench V3 - Banking | 38.1 | 15.3 |
| Terminal-Bench 2.1 | 88 | 47.6 |
| Terminal-Bench Hard | 62.9 | 31.1 |
| WideSearch | 81.2 | 60.5 |
| τ²-Bench Telecom (AA run) | 98.5 | 93.6 |
| τ³-Bench | 26.8 | 13.6 |
Best tracked score per model per benchmark (default configuration; source-attributed). ↓ marks lower-is-better metrics. Open either model for its full surface, provenance and pricing. Quoted rates are the price-setter row we currently track for each model — its direct or vendor-official listing where one exists (direct, direct), otherwise the lowest tracked offer.