OpenVLM MMT-Bench - Set-of-Mark Recognition Leaderboard — The Aggregate

OpenVLM MMT-Bench - Set-of-Mark Recognition — leaderboard

Metric: Score (%). Source: huggingface.co. 207 models tracked.

Top models

#	Model	Score
1	GPT-4.1 Mini	100
2	InternVL3-38B	100
3	InternVL3-14B	100
4	GPT-4.1 (2025-04-14)	100
5	InternVL3-78B	94.7
6	InternVL3-8B	94.7
7	Qwen 2 VL 72B	94.7
8	Llama 3.2 11B Instruct	94.7
9	Claude 3.5 Sonnet	94.7
10	Llama 3.2 90B Vision Instruct	94.7
11	Qwen 2 VL 7B	89.5
12	Claude 3.7 Sonnet	89.5
13	Gemini 1.5 Pro	89.5
14	Gemini 1.5 Flash	89.5
15	MiniCPM-V-2.6	84.2

Interactive version: theaggregate.ai/benchmark?slug=openvlm-mmt-bench-set-of-mark-recognition · How the rankings work · Data refreshed daily, snapshot 2026-07-22.