| Benchmark | Claude Fable 5 | Llama 3.1 Instruct 405B |
|---|---|---|
| AA Agentic Index | 56.6 | 6.3 |
| AA Intelligence | 62.1 | 8.5 |
| AA-LCR | 76.7 | 24.7 |
| AA-Omniscience | 43.3 | -17.1 |
| Artificial Analysis Coding Index | 76.5 | 14.5 |
| critpt | 28.6 | 0 |
| gdpval | 61.9 | 0 |
| GPQA Diamond | 92.6 | 51.5 |
| HLE | 64.5 | 4.2 |
| IFBench | 63.5 | 39 |
| OmniScience Accuracy | 65.3 | 23.2 |
| OmniScience Non-Hallucination | 36.4 | 49 |
| scicode | 60.2 | 29.9 |
| simplebench | 81.9 | 23 |
| SWE-bench Pro | 80 | 11.2 |
| SWE-bench Verified | 95 | 24.5 |
| Terminal-Bench Hard | 62.9 | 6.8 |
| τ²-Bench Telecom (AA run) | 98.5 | 19 |
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.