Colon-Bench - Lesion Classification F1: leaderboard
Metric: F1 (%) of the lesion-present class when classifying Colon-Bench's 790 colonoscopy clips (272 with a confirmed lesion, 518 lesion-free) from the video clip; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 17 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Flash Lite | 80.7 | #174 |
| 2 | GPT-5.4 | 77.8 | #76 |
| 3 | Gemini 3 Pro | 77.3 | #77 |
| 4 | GPT-5.2 | 76.7 | #105 |
| 5 | Gemini 3 Flash | 70.5 | #93 |
| 6 | Seed-1.6 | 69.4 | #257 |
| 7 | GLM-4.6V | 62.2 | #309 |
| 8 | Gemini 2.5 Flash Lite | 58.7 | #413 |
| 9 | Molmo2-8B | 49.6 | #522 |
| 10 | Seed 1.6 Flash | 38.6 | #427 |
| 11 | Qwen 3.5 Plus | 29.3 | #123 |
| 12 | Qwen 3 VL 235B A22B | 1.4 | #228 |
| 13 | Qwen 3.5 397B A17B | 0.7 | #141 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=colon-bench-lesion-classification-f1 · How It Works · Data refreshed daily, snapshot 2026-10-11.