Benchmark
SuperGPQA
285 graduate disciplines; breadth comes at the cost of per-field sample size.
- Released
- 2025-02-20
- Categories
- knowledge
- Openness
- unknown
- Source
- model_reports
- Reported scores
- 1
Reported scores
model_reports
| Model | Organization | Reported value | Reported | Evidence |
|---|---|---|---|---|
| Qwen3.5-397B-A17B | Qwen | 70.4 | 2026-02-16 | evidence |
Scores are partitioned by the source that reported them and are never merged into a single cross-source ranking, because the sources measure different things and say so.