| # | Model | Vendor | Best score | Runs | Last seen |
|---|---|---|---|---|---|
| 1 | Gemma 4 E2B | 64.5 | 1 | 2026-08-12 | |
| 2 | LFM2.5-VL-3B | Liquid AI | 60.6 | 3 | 2026-08-23 |
| 3 | LFM2.5-VL-1.6B | Liquid AI | 52.3 | 1 | 2026-08-09 |
| 4 | LFM2-VL-3B | Liquid AI | 51.8 | 2 | 2026-08-22 |
| 5 | LFM2-VL-3B (3.1B) | Liquid AI | 51.4 | 1 | 2026-08-12 |
| 6 | InternVL3_5-2B | OpenGVLab | 47.3 | 1 | 2026-08-09 |
| 7 | LFM2-VL-1.6B | Liquid AI | 46.4 | 1 | 2026-08-09 |
| 8 | LFM2.5-VL-450M-Extract | Liquid AI | 45 | 1 | 2026-08-09 |
| 9 | Qwen2.5 VL 3B | Alibaba | 38.6 | 1 | 2026-08-09 |
| 10 | InternVL3-2B | OpenGVLab | 38.5 | 1 | 2026-08-09 |
| 11 | InternVL3.5-1B | OpenGVLab | 36.2 | 1 | 2026-08-09 |
| 12 | LFM2-VL-450M | Liquid AI | 33.1 | 1 | 2026-08-09 |
Best tracked score per model (default configuration; source-attributed and verification-tiered). Open a model for its full benchmark surface, provenance and pricing.