MedRepBench - Item Name Recall (OCR Text Only): leaderboard

Metric: Field-level recall (%; item name of each ground-truth lab item, matched by exact normalized item name within the model's top-K items, K = number of reference items; text LLM on the OCR text). Source: arxiv.org. Saturation forecast: Estimated already saturated. 12 models tracked.

Top models

#ModelScore
1DeepSeek V396.48
2Qwen 3 235B A22B 250796.42
3Qwen 3 30B A3B96.08
4Qwen 3 14B95.75
5Qwen 3 32B95.66
6DeepSeek R1 Distill Qwen 14B95.5
7Qwen 3 235B A22B94.96
8Qwen 3 8B94.93
9Qwen 3 4B94.64
10DeepSeek R1 Distill Qwen 32B92.93
11Qwen 3 1.7B92.88
12DeepSeek-R1-Distill-Qwen-7B91.68

Interactive version: theaggregate.ai/benchmark?slug=medrepbench-item-name-recall-ocr-text-only · How It Works · Data refreshed daily, snapshot 2026-09-26.