| Benchmark | Llama 3.1 Instruct 405B | Qwen3.7 Plus Preview |
|---|---|---|
| AA Agentic Index | 6.3 | 20.8 |
| AA Intelligence | 8.5 | 39.4 |
| AA-LCR | 24.7 | 69 |
| AA-Omniscience | -17.1 | 2.4 |
| Artificial Analysis Coding Index | 14.5 | 55.9 |
| critpt | 0 | 9.1 |
| gdpval | 0 | 22.2 |
| GPQA Diamond | 51.5 | 90.3 |
| HLE | 4.2 | 35.6 |
| IFBench | 39 | 79.1 |
| ifeval | 88.6 | 94.6 |
| mmlu_redux | 86.2 | 94.5 |
| MMLU-Pro | 73.4 | 88.5 |
| mmmlu | 73.8 | 89 |
| OmniScience Accuracy | 23.2 | 22.5 |
| OmniScience Non-Hallucination | 49 | 74.5 |
| scicode | 29.9 | 51.3 |
| SWE-bench Pro | 11.2 | 57.6 |
| SWE-bench Verified | 24.5 | 77.7 |
| Terminal-Bench Hard | 6.8 | 47 |
| τ²-Bench Telecom (AA run) | 19 | 93 |
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.