SEA-Vision - TEC-VQA: leaderboard
Metric: Accuracy (%) over all languages, weighted by question count, on SEA-Vision TEC-VQA (text-centric visual question answering on scene and document images, 7,496 question-answer pairs across 11 Southeast Asian languages, questions answered in the image's language, zero-shot); an answer is correct when it contains at least one reference answer; higher is better. Source: arxiv.org. Saturation forecast: Around 2029. 12 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3 VL 32B Instruct | 40.14 | #276 |
| 2 | Gemini 2.5 Pro | 39.49 | #145 |
| 3 | Qwen 2.5 VL 72B Instruct | 37.43 | #364 |
| 4 | GPT-4o | 35.49 | #333 |
| 5 | InternVL3-78B | 30.94 | #345 |
| 6 | MiniCPM-V-2.6 | 26.16 | #825 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=sea-vision-tec-vqa · How It Works · Data refreshed daily, snapshot 2026-10-11.