WorldVQA F-score
Multimodal · 2026-01-28
WorldVQA F-score is the harmonic mean of Accuracy and Correct Given Attempted across the first eight categories, penalizing both abstention and incorrect answers.
Top models (higher is better)
| Model | Score |
|---|---|
| Gemini 3 Pro Preview | 47.5 |
| Kimi K2.5 | 46.8 |
| Opus 4.5 | 37.5 |
| Gemini 2.5 Pro | 36.9 |
| Seed1.5-VL | 35.2 |
| GPT-5.2 | 28.7 |
| GPT-5.1 | 26.7 |
| Qwen3-VL-235B-A22B-Instruct | 23.5 |
| GPT-4o | 23.3 |
| Grok 4.1 Fast | 21.1 |
| Sonnet 4.5 | 20.9 |
| GLM-4.6V | 19.0 |
| Grok 4 Fast | 18.9 |
| Qwen3 VL 32B Instruct | 17.7 |
| GLM 4.6V Flash | 14.8 |