| # | Model | Vendor | Best score | Runs | Last seen |
|---|---|---|---|---|---|
| 1 | Mage-VL-4B | Microsoft | 95.1 | 1 | 2026-07-26 |
| 2 | Qwen3 VL 4B (Reasoning) | Alibaba | 94.7 | 1 | 2026-07-26 |
| 3 | Phi 4 MM 5.6B | Microsoft | 92.8 | 1 | 2026-07-26 |
| 4 | Phi 4 R V 15B | Microsoft | 76.2 | 1 | 2026-07-26 |
| 5 | Qwen3.5 2B | Alibaba | 0.9 | 2 | 2026-08-12 |
| 6 | North-Micro-Vision-Instruct | Cohere | 0.9 | 2 | 2026-08-12 |
| 7 | Ministral 3 3B | Mistral | 0.9 | 2 | 2026-08-12 |
| 8 | LFM2.5-VL-1.6B | Liquid AI | 0.9 | 2 | 2026-08-12 |
| 9 | Phi 3.5 Vision Instruct | Microsoft | 0.9 | 2 | 2026-08-12 |
| 10 | Qwen3 VL 2B Instruct | Alibaba | 0.8 | 2 | 2026-08-12 |
| 11 | Gemma 4 E2B | 0.7 | 2 | 2026-08-12 |
Best tracked score per model (default configuration; source-attributed and verification-tiered). Open a model for its full benchmark surface, provenance and pricing.