| Benchmark | Claude Opus 4.5 | GLM-5.1 |
|---|---|---|
| AA Agentic Index | 59.6 | 66 |
| AA Intelligence | 41.9 | 41 |
| AA-LCR | 76 | 68 |
| AA-Omniscience | 14 | 0.8 |
| aime_2026 | 95.1 | 95.8 |
| arena_elo | 1473 | 1468 |
| arena_text_factuality | 1458 | 1458 |
| Artificial Analysis Coding Index | 47.8 | 55.8 |
| browsecomp | 37 | 79.3 |
| browsecomp_with_context_manager | 67.8 | 79.3 |
| coding_arena_elo | 1495 | 1510 |
| critpt | 4.6 | 4.6 |
| cybergym | 50.6 | 68.7 |
| frontiermath_tier_4 | 4.2 | 12.5 |
| gdpval | 47.3 | 49.5 |
| GPQA Diamond | 87 | 86.8 |
| HLE | 30.8 | 52.3 |
| HLE (with tools) | 43.4 | 52.3 |
| hmmt_feb_2026 | 85.3 | 89.4 |
| hmmt_nov_2025 | 93.3 | 94 |
| IFBench | 58 | 76.3 |
| imo_answer_bench | 84 | 83.8 |
| livebench | 76 | 70.2 |
| MCP Atlas | 69.8 | 75.6 |
| MMLU-Pro | 90 | 86 |
| nl2repo | 43.2 | 42.7 |
| OmniScience Accuracy | 46.6 | 25.2 |
| OmniScience Non-Hallucination | 39 | 70.1 |
| scicode | 50 | 43.8 |
| simplebench | 62 | 55.1 |
| simpleqa_verified | 41.8 | 38.1 |
| SWE-bench Multilingual | 77.5 | 73.3 |
| SWE-bench Pro | 57.1 | 58.4 |
| Terminal-Bench 2.0 | 59.3 | 69 |
| Terminal-Bench Hard | 47 | 43.2 |
| tool_decathlon | 43.5 | 40.7 |
| toolathlon | 43.5 | 40.7 |
| τ²-Bench Telecom (AA run) | 89.5 | 97.7 |
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.