| Benchmark | GLM-5.2 Full Open Source | Qwen3.5 397B A17B |
|---|---|---|
| AA Agentic Index | 45.7 | 53.3 |
| AA Intelligence | 52.6 | 34.3 |
| AA-LCR | 76.7 | 72.7 |
| AA-Omniscience | 4.4 | -30.8 |
| aime_2026 | 99.2 | 94.2 |
| apexAgents | 33.7 | 15.3 |
| arena_elo | 1471 | 1442 |
| arena_text_factuality | 1460 | 1445 |
| Artificial Analysis Coding Index | 68.8 | 48.2 |
| coding_arena_elo | 1582 | 1399 |
| critpt | 20.9 | 1.7 |
| gdpval | 50.3 | 35.8 |
| GDPval-AA v2 | 1514 | 962 |
| Global-MMLU-Lite | 89.2 | 90 |
| GPQA Diamond | 91.2 | 89.3 |
| HLE | 54.7 | 29 |
| HLE (with tools) | 54.7 | 48.3 |
| HMMT 2025 | 94.4 | 92.7 |
| hmmt_feb_2026 | 92.5 | 87.9 |
| hmmt_nov_2025 | 94.4 | 92.7 |
| IFBench | 73.3 | 78.8 |
| imo_answer_bench | 91 | 80.9 |
| itbenchSre | 42.7 | 34.1 |
| nl2repo | 48.9 | 32.2 |
| OmniScience Accuracy | 24.3 | 30.8 |
| OmniScience Non-Hallucination | 73.7 | 17.3 |
| scicode | 50.5 | 42 |
| simpleqa_verified | 38.1 | 26 |
| strongreject | 98.5 | 99.4 |
| SWE-bench Pro | 62.1 | 50.9 |
| Tau 3 Banking | 26.8 | 13.4 |
| TauBench V3 - Banking | 34.6 | 13.4 |
| Terminal-Bench 2.1 | 82.7 | 51.3 |
| Terminal-Bench Hard | 50.8 | 40.9 |
| toolathlon | 48.2 | 38.3 |
| τ²-Bench Telecom (AA run) | 99.1 | 95.6 |
| τ³-Bench | 26.8 | 13.4 |
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.