FlagEval VQA - Text Understanding (English): leaderboard
BAAI FlagEval’s multimodal VQA leaderboard, Text Understanding (English) slice: reading and reasoning over English text inside images.
Metric: Score. Source: flageval.baai.ac.cn. Saturation forecast: Around December 2026. 24 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 81.5 | #152 |
| 2 | Gemini 2.5 Flash (Thinking) | 80.8 | #245 (Gemini 2.5 Flash) |
| 3 | Gemini 2.5 Flash (Non-reasoning) | 77.1 | #245 (Gemini 2.5 Flash) |
| 4 | GPT-4.1 | 74.6 | #248 |
| 5 | GPT-5 (High) | 74.4 | #98 (GPT-5) |
| 6 | GPT-5 (Low) | 73 | #98 (GPT-5) |
| 7 | O3 Mini (High) | 72.4 | #272 (O3 Mini) |
| 8 | ERNIE-4.5-VL-424B-A47B | 72.3 | #391 |
| 9 | O3 Mini (Low) | 71.5 | #272 (O3 Mini) |
| 10 | GLM-4.5V | 67.5 | #346 |
| 11 | O4 Mini (High) | 65.3 | #180 (O4 Mini) |
| 12 | O4 Mini (Low) | 63.9 | #180 (O4 Mini) |
| 13 | Mistral Medium 3 | 60.6 | #500 |
| 14 | Mistral Medium 3.1 | 60.4 | #408 |
| 15 | Step-3 | 54.9 | #365 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=flageval-vqa-text-understanding-english · How It Works · Data refreshed daily, snapshot 2026-10-20.