QCalEval (In-Context) - Calibration Diagnosis: leaderboard
Metric: Accuracy (%) of the family-specific calibration status code, with one labeled demonstration per scenario type, on the QCalEval quantum-calibration plots with in-context demonstrations from the same experiment family (scenario types with a single sample left out), greedy decoding; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 17 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 89.8 |
| 2 | Claude Opus 4.6 | 89.4 |
| 3 | Gemma 4 31B (IT) | 86 |
| 4 | Gemini 3.1 Flash Lite | 82.2 |
| 5 | GPT-5.4 | 81.4 |
| 6 | Claude Sonnet 4.6 | 78 |
| 7 | Claude Haiku 4.5 | 73.1 |
| 8 | GPT-5.4 Mini | 66.9 |
| 9 | InternVL3-38B | 55.1 |
| 10 | Qwen 3.5 27B | 45.8 |
| 11 | InternVL3-78B | 44.3 |
| 12 | Qwen 3.5 397B A17B | 42.4 |
| 13 | Qwen 3.5 122B A10B | 35.2 |
| 14 | Qwen 3.5 9B | 33.9 |
| 15 | Qwen 3.5 35B A3B | 33.9 |
Interactive version: theaggregate.ai/benchmark?slug=qcaleval-in-context-calibration-diagnosis · How It Works · Data refreshed daily, snapshot 2026-10-07.