InfographicVQA — Kimi-K2.5-sourced val score; diff source/score pop than InfoVQA (Acc) - kept separate.
| # | Model | Vendor | Best score | Runs | Last seen |
|---|---|---|---|---|---|
| 1 | Kimi K2.5 | Moonshot | 92.6 | 1 | 2026-06-15 |
| 2 | Qwen3 VL 235B A22B Reasoning | Alibaba | 89.5 | 1 | 2026-06-15 |
| 3 | GPT-5.2 | OpenAI | 84 | 1 | 2026-06-15 |
| 4 | Mage-VL-4B | Microsoft | 80.3 | 1 | 2026-07-26 |
| 5 | Qwen3 VL 4B (Reasoning) | Alibaba | 79.5 | 1 | 2026-07-26 |
| 6 | Claude Opus 4.5 | Anthropic | 76.9 | 1 | 2026-06-15 |
| 7 | Phi 4 MM 5.6B | Microsoft | 71.8 | 1 | 2026-07-26 |
| 8 | LFM2.5-VL-1.6B | Liquid AI | 62.7 | 1 | 2026-08-09 |
| 9 | InternVL3.5-1B | OpenGVLab | 61 | 1 | 2026-08-09 |
| 10 | LFM2-VL-1.6B | Liquid AI | 58.4 | 1 | 2026-08-09 |
| 11 | Gemini 3 Pro | 57.2 | 1 | 2026-06-15 | |
| 12 | Phi 4 R V 15B | Microsoft | 55.4 | 1 | 2026-07-26 |
| 13 | LFM2-VL-450M | Liquid AI | 44.6 | 1 | 2026-08-09 |
| 14 | LFM2.5-VL-450M-Extract | Liquid AI | 43 | 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.