QCalEval - Outcome Classification: leaderboard
Metric: Accuracy (%) of the four-way experimental outcome label (expected behavior, suboptimal parameters, anomalous behavior, apparatus issue), on the 243 quantum-calibration plots of QCalEval (22 experiment families, superconducting qubits and neutral atoms), zero-shot with the family background text, greedy decoding; higher is better. Source: arxiv.org. Saturation forecast: Around August 2027. 18 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 57.2 |
| 2 | Gemma 4 31B (IT) | 54.3 |
| 3 | Gemini 3.1 Flash Lite | 53.5 |
| 4 | GPT-5.4 | 52.7 |
| 5 | Claude Opus 4.6 | 49 |
| 6 | Claude Sonnet 4.6 | 48.6 |
| 7 | Qwen 3.5 27B | 45.7 |
| 8 | Qwen 3.5 122B A10B | 44 |
| 9 | Qwen 3.5 397B A17B | 42.8 |
| 10 | Qwen 3.5 35B A3B | 39.9 |
| 11 | GPT-5.4 Mini | 39.5 |
| 12 | Qwen 3.5 9B | 37.9 |
| 13 | InternVL3-78B | 37 |
| 14 | Claude Haiku 4.5 | 36.6 |
| 15 | InternVL3-38B | 34.6 |
Interactive version: theaggregate.ai/benchmark?slug=qcaleval-outcome-classification · How It Works · Data refreshed daily, snapshot 2026-10-07.