GB/T-Bench - Direct Review Diagnosis Recall: leaderboard
Metric: Diagnosis recall (%; exact section, dimension and error-type match). Source: arxiv.org. 14 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.6 Sol | 52.03 |
| 2 | GPT-5.5 | 50.52 |
| 3 | Claude Fable 5 | 49.93 |
| 4 | Qwen 3.7 Plus | 43.51 |
| 5 | GPT-5.4 | 42.87 |
| 6 | Gemini 3.5 Flash | 40.87 |
| 7 | DeepSeek V4 Pro | 39.61 |
| 8 | GPT-5.3 | 36.2 |
| 9 | DeepSeek V4 Flash | 34.56 |
| 10 | GPT-5.4 Mini | 30.52 |
| 11 | Claude Opus 4.8 | 29.04 |
| 12 | Qwen 2.5 VL 72B Instruct | 27.31 |
| 13 | Qwen 3 235B A22B 2507 Instruct | 25.51 |
| 14 | MiniMax-M2.7 | 25.16 |
Interactive version: theaggregate.ai/benchmark?slug=gb-t-bench-direct-review-diagnosis-recall · How It Works · Data refreshed daily, snapshot 2026-09-19.