Fine-grained visual understanding on MME RealWorld
65.3ScoreSwimBird
Evaluation results
| Method | Links | |
|---|---|---|
| SwimBirdBackbone=Qwen3-VL 8B2026.02 | 65.3 | |
| DeepEyesV2Model Category=Multimodal Agentic Models2026.02 | 64.9 | |
| ThymeModel Category=Multimodal Agentic Models2026.02 | 64.8 | |
| Pixel ReasonerModel Category=Multimodal Agentic Models2026.02 | 64.4 | |
| DeepEyesModel Category=Multimodal Agentic Models2026.02 | 64.1 | |
| GPT-4oModel Category=Textual Reasoning Models2026.02 | 62.8 | |
| Qwen3-VL-8B-InstructModel Category=Textual Reasoning Models, Reproduced=true2026.02 | 61.9 | |
| Qwen2.5-VL-32B-InstructModel Category=Textual Reasoning Models2026.02 | 59.1 | |
| LLaVA-OneVisonModel Category=Textual Reasoning Models2026.02 | 57.4 | |
| Qwen2.5-VL-7B-InstructModel Category=Textual Reasoning Models2026.02 | 56.8 |