GB/T-Bench - Diagnosis Recall - Terminology: leaderboard
Metric: Diagnosis recall (%; exact section, dimension and error-type match). Source: arxiv.org. 14 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.5 Flash | 25.97 |
| 2 | GPT-5.4 Mini | 25.4 |
| 3 | Claude Fable 5 | 24.77 |
| 4 | GPT-5.6 Sol | 22.94 |
| 5 | Qwen 3.7 Plus | 21.68 |
| 6 | GPT-5.3 | 20.15 |
| 7 | Qwen 3 235B A22B 2507 Instruct | 19.34 |
| 8 | GPT-5.5 | 19.09 |
| 9 | DeepSeek V4 Pro | 18.2 |
| 10 | Qwen 2.5 VL 72B Instruct | 18.2 |
| 11 | GPT-5.4 | 17.72 |
| 12 | DeepSeek V4 Flash | 16.26 |
| 13 | Claude Opus 4.8 | 13.19 |
| 14 | MiniMax-M2.7 | 11.97 |
Interactive version: theaggregate.ai/benchmark?slug=gb-t-bench-diagnosis-recall-terminology · How It Works · Data refreshed daily, snapshot 2026-09-19.