Question Answering on Real-World Dataset A
88.9Scenario 1 ScoreKairosVL (7B)
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| KairosVL (7B)Open-Source=true2026.02 | 88.9 | 70.4 | 73.3 | |
| Qwen2.5VL-72BOpen-Source=true2026.02 | 87.8 | 60 | 70.7 | |
| GPT-4oOpen-Source=false2026.02 | 83.3 | 64.5 | 77.3 | |
| KairosVL-SFT (7B)2026.02 | 81.8 | 60.9 | 58.2 | |
| Qwen2.5VL-32BOpen-Source=true2026.02 | 79.6 | 56.8 | 46.7 | |
| Gemini-2.5FlashOpen-Source=false2026.02 | 79.4 | 62.5 | 48.7 | |
| Qwen2.5VL-7BOpen-Source=true2026.02 | 75.6 | 53.3 | 45.3 | |
| GLM4V-PlusOpen-Source=false2026.02 | 66.7 | 70 | 49.3 | |
| InternVL3-8BOpen-Source=true2026.02 | 57.4 | 53 | 62 |