| Benchmark | GLM-5.2 Full Open Source | Nemotron 3 Super 120B A12B |
|---|---|---|
| AA Agentic Index | 45.7 | 8.8 |
| AA Intelligence | 52.6 | 25.7 |
| AA-LCR | 76.7 | 60.3 |
| AA-Omniscience | 4.4 | -41.5 |
| aime_2026 | 99.2 | 91.7 |
| apexAgents | 33.7 | 1.8 |
| Artificial Analysis Coding Index | 68.8 | 37.7 |
| critpt | 20.9 | 3.1 |
| gdpval | 50.3 | 9.9 |
| GPQA Diamond | 91.2 | 82.7 |
| HLE | 54.7 | 22.8 |
| HMMT 2025 | 94.4 | 94.7 |
| hmmt_feb_2026 | 92.5 | 85.2 |
| IFBench | 73.3 | 72.6 |
| itbenchSre | 42.7 | 1.1 |
| OmniScience Accuracy | 24.3 | 24.3 |
| OmniScience Non-Hallucination | 73.7 | 13 |
| scicode | 50.5 | 42 |
| TauBench V3 - Banking | 34.6 | 10.3 |
| Terminal-Bench 2.1 | 82.7 | 38.6 |
| Terminal-Bench Hard | 50.8 | 28.8 |
| τ²-Bench Telecom (AA run) | 99.1 | 67.8 |
| τ³-Bench | 26.8 | 10.1 |
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.