MI-CXR - Trajectory Summary (Single Abnormality): leaderboard

Trajectory Summary (Single Abnormality) questions of MI-CXR (2026), five-option questions over five-visit chest X-ray timelines from MIMIC-CXR, answered zero-shot from all five images: summarize the course of one abnormality across all visits. Accuracy in percent (chance 20); higher is better.

Metric: Global Trajectory Summarization for one abnormality across all visits, five-option multiple-choice accuracy (%, times 100; chance 20) on MI-CXR, longitudinal reasoning over five-visit chest X-ray timelines from MIMIC-CXR (5,311 questions), zero-shot single-step prompting with all five images, temperature 0; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 14 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1GPT-5.255.8#185
2Gemini 3 Pro (Preview)55.6#119
3InternVL3.5-38B51#1021
4Claude Sonnet 4.538.9#261
5InternVL3.5-8B38.9#1604
6Lingshu-32B38.8#1113
7InternVL3.5-14B37.4#1504
8Qwen 3 VL 32B36.3#622
9Qwen 3 VL 8B32.8#917
10Lingshu-7B32.4#1635
11IDEFICS2-8B28.1#5013
12MedGemma-4B25.9#2557
13MedGemma-27B-IT25.5#1350
14MI-CXR DeepSeek-VL-16B (checkpoint unspecified)16

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=mi-cxr-trajectory-summary-single-abnormality · How It Works · Data refreshed daily, snapshot 2026-11-21.