OpenVLM COCO Captions — leaderboard
OpenCompass OpenVLM COCO captioning results: image caption generation on MS COCO validation images, scored with automatic caption similarity metrics such as CIDEr.
Metric: CIDEr. Source: huggingface.co. Status: saturation imminent. 211 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 1.0 Pro | 9.7 |
| 2 | InternVL3-14B | 2.1 |
| 3 | InternVL3-8B | 1.7 |
| 4 | InternVL3-78B | 0.9 |
| 5 | InternVL3-38B | 0.8 |
| 6 | GPT-4.1 Nano | 0.4 |
| 7 | InternVL2-8B | 0.3 |
| 8 | GPT-4.1 Mini | 0.1 |
| 9 | Grok 2 (1212) | 0.1 |
| 10 | Gemini 1.5 Flash | 0.1 |
| 11 | Gemma 3 4B | 0.1 |
| 12 | Qwen 2 VL 72B | 0 |
| 13 | Qwen 2 VL 7B | 0 |
| 14 | Qwen 2 VL 2B | 0 |
| 15 | Ovis2-8B | 0 |
Interactive version: theaggregate.ai/benchmark?slug=openvlm-coco-captions · How the rankings work · Data refreshed daily, snapshot 2026-07-22.