| Benchmark | Claude Opus 4.8 | Qwen3.7 Plus Preview |
|---|---|---|
| AA Agentic Index | 49.4 | 20.8 |
| AA Intelligence | 57.3 | 39.4 |
| AA-LCR | 73 | 69 |
| AA-Omniscience | 28.8 | 2.4 |
| Agents' Last Exam | 27 | 33.6 |
| arena_elo | 1482 | 1458 |
| arena_text_factuality | 1464 | 1454 |
| arena_vision | 1280 | 1263 |
| Artificial Analysis Coding Index | 74.3 | 55.9 |
| charxiv_rq | 80.5 | 85.9 |
| critpt | 20.9 | 9.1 |
| DeepSWE 1.1 | 59 | 14.2 |
| Finance Agent v2 | 53.9 | 38.2 |
| gdpval | 54.2 | 22.2 |
| GPQA Diamond | 93.6 | 90.3 |
| HLE | 57.9 | 35.6 |
| hmmt_feb_2026 | 96.7 | 92.9 |
| IFBench | 62.2 | 79.1 |
| imo_answer_bench | 83.5 | 86 |
| include | 87.6 | 83 |
| JobBench | 48.4 | 27.6 |
| mathvision | 86.7 | 90.3 |
| MCP Atlas | 83.6 | 73.2 |
| MMMU-Pro | 78.9 | 80.5 |
| nl2repo | 69.7 | 41.1 |
| OmniScience Accuracy | 48.8 | 22.5 |
| OmniScience Non-Hallucination | 60.7 | 74.5 |
| OSWorld-Verified | 83.4 | 73.3 |
| scicode | 53.5 | 51.3 |
| screenspot_pro_no_tools | 87.9 | 79 |
| SWE-bench Multilingual | 84.4 | 75.8 |
| SWE-bench Pro | 69.2 | 57.6 |
| SWE-bench Verified | 88.6 | 77.7 |
| TauBench V3 - Banking | 34.2 | 17.9 |
| Terminal-Bench 2.0 | 74.6 | 70.3 |
| Terminal-Bench 2.1 | 85 | 61 |
| Terminal-Bench Hard | 58.3 | 47 |
| Video-MME | 86 | 88 |
| τ²-Bench Telecom (AA run) | 94.4 | 93 |
| τ³-Bench | 27.6 | 17.9 |
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.