OCR-related Understanding Tasks on SEED-Bench-2-Plus
76.5AccuracyQwen2.5-VL-7B AIF
Evaluation results
| Method | Links | |
|---|---|---|
| Qwen2.5-VL-7B AIFmethod=Adaptive Information Flow2026.04 | 76.5 | |
| Qwen2.5-VL-72BModel Size=72B2025.02 | 73 | |
| Qwen2.5-VL-72B2026.04 | 73 | |
| MetisBackbone=Qwen3-VL-8B-Instruct2026.04 | 72.5 | |
| Qwen2.5-VL-32B-InstructModel Category=Open-Source Models2026.04 | 72.4 | |
| GPT-4o2025.02 | 72 | |
| GPT-4o2026.04 | 72 | |
| Claude-3.5 Sonnet2025.02 | 71.7 | |
| Claude-3.5 Sonnet2026.04 | 71.7 | |
| InternVL2.5-78BModel Size=78B2025.02 | 71.3 | |
| Qwen3-VL-8B-InstructModel Category=Open-Source Models2026.04 | 71 | |
| Gemini 1.5 Pro2025.02 | 70.8 | |
| DeepEyesV2Model Category=Agentic Multimodal Models2026.04 | 70.5 | |
| Qwen2.5-VL-7BModel Size=7B2025.02 | 70.4 | |
| Qwen2.5-VL-7B-InstructModel Category=Open-Source Models2026.04 | 70.4 | |
| Qwen2.5-VL-7B2026.04 | 70.4 | |
| InternVL3-8BModel Category=Open-Source Models2026.04 | 69.7 | |
| Qwen2.5-VL-3BModel Size=3B2025.02 | 67.6 | |
| LLaVA-OneVisionModel Category=Open-Source Models2026.04 | 65.4 | |
| LLaVA-1.5-7B AIFmethod=Adaptive Information Flow2026.04 | 44.9 | |
| LLaVA-1.5-7B2026.04 | 41.3 |