Calibration
SuperGPQA → HLE
HLE is estimated from SuperGPQA with a offset logistic curve fitted on 14 models measured on both: y = 0.0585 + (0.5113 − 0.0585) / (1 + exp(−19.17·(x − 0.6886))), R² = 0.94, cross-validated error 4.8 pp. It is used for 8 estimates.
| Estimated model | SuperGPQA | HLE | Source |
|---|---|---|---|
| Qwen 3.6 Max (preview) | 73.9% | 38.6% | estimated ± 4.8 pp, high confidence |
| Seed 2.1 Pro | 70.8% | 32.6% | estimated ± 4.8 pp, high confidence |
| Seed 2.1 Turbo | 67.4% | 25.3% | estimated ± 4.8 pp, high confidence |
| Qwen3.5-122B-A10B | 67.1% | 24.7% | estimated ± 4.8 pp, high confidence |
| Qwen3.5-27B | 65.6% | 21.6% | estimated ± 4.8 pp, high confidence |
| Qwen3.5-35B-A3B | 63.4% | 17.6% | estimated ± 4.8 pp, high confidence |
| Qwen3 235B 2507 | 62.6% | 16.3% | estimated ± 4.8 pp, high confidence |
| MiniCPM5-1B | 23.1% | 5.9% | estimated ± 4.8 pp, medium confidence |
Fitted on
| Model | SuperGPQA | HLE |
|---|---|---|
| Claude Opus 4.5 | 70.6% | 30.8% |
| Claude Opus 4.6 | 95.0% | 53.0% |
| Claude Sonnet 4.6 | 95.0% | 49.0% |
| GLM-5 | 66.8% | 30.5% |
| Kimi K2 | 57.2% | 4.7% |
| Kimi K2.5 | 69.2% | 30.1% |
| MiniCPM5-2B | 40.8% | 8.9% |
| Qwen3 235B 2507 (Reasoning) | 64.9% | 18.2% |
| Qwen3.5 397B | 70.4% | 28.7% |
| Qwen3.6-27B | 66.0% | 24.0% |
| Qwen3.6-35B-A3B | 64.7% | 21.4% |
| Qwen3.6 Plus | 71.6% | 28.8% |
| Qwen3.7 Max | 73.6% | 41.4% |
| Qwen3.7 Plus | 71.4% | 34.7% |