Hallucination Evaluation on MVH-Bench Overall
489.89MVH ScoreQwen2.5-VL-7B
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen2.5-VL-7BSource Type=Open-Source2026.03 | 489.89 | |
| Gemini 2.5 FlashSource Type=Closed-Source2026.03 | 486.64 | |
| GPT-4oSource Type=Closed-Source2026.03 | 478.32 | |
| Gemini 2 FlashSource Type=Closed-Source2026.03 | 477.35 | |
| Claude 4.5 SonnetSource Type=Closed-Source2026.03 | 471.68 | |
| GPT-4o miniSource Type=Closed-Source2026.03 | 424.99 | |
| Qwen2-VL-7BSource Type=Open-Source2026.03 | 413.14 | |
| InternVL3-14BSource Type=Open-Source2026.03 | 411.65 | |
| LLaVA-NeXT-I-7BSource Type=Open-Source2026.03 | 402.87 | |
| InternVL3-8BSource Type=Open-Source2026.03 | 379.39 | |
| Mantis-8B-Idefics2Source Type=Open-Source2026.03 | 375.76 | |
| LLaVA-OneVision-7BSource Type=Open-Source2026.03 | 373.49 | |
| Gemma3-12BSource Type=Open-Source2026.03 | 348.23 | |
| Deepseek-VL-Chat-7BSource Type=Open-Source2026.03 | 312.24 | |
| Llama-3.2-11B-V-ISource Type=Open-Source2026.03 | 306.24 |