Multimodal Perception on HR-4K
87.88AccuracyGemini-3-Flash
Evaluation results
| Method | Links | |
|---|---|---|
| Gemini-3-Flash2026.02 | 87.88 | |
| Qwen3-VL-235B-A22Bnumber of parameters=235B2026.02 | 84.5 | |
| ZwZ-8Bnumber of parameters=8B2026.02 | 84.38 | |
| Gemini-2.5-ProModel Category=Closed-source models2026.06 | 83.9 | |
| TACO (Ours-7B)Model Category=Code-tool / visual-agent models2026.06 | 83.8 | |
| Kimi-K2.5 (1T)number of parameters=1T2026.02 | 81.87 | |
| ZwZ-4Bnumber of parameters=4B2026.02 | 81.75 | |
| GLM-4.5V (108B)number of parameters=108B2026.02 | 81.63 | |
| Qwen3-VL-8BModel Category=Open-source MLLMs (no visual tool), Instruct variant=true2026.06 | 78.9 | |
| Qwen3-VL-8Bnumber of parameters=8B2026.02 | 78.88 | |
| Qwen3-VL-4Bnumber of parameters=4B2026.02 | 78.25 | |
| PyVision-RL-7BModel Category=Code-tool / visual-agent models2026.06 | 78.1 | |
| DeepEyes-v2-7BModel Category=Code-tool / visual-agent models2026.06 | 77.9 | |
| Mini-o3-7BModel Category=Code-tool / visual-agent models2026.06 | 77.5 | |
| Thyme-7BModel Category=Code-tool / visual-agent models2026.06 | 77 | |
| CodeV-RL-7BModel Category=Code-tool / visual-agent models2026.06 | 76.1 | |
| ZwZ-7Bnumber of parameters=7B2026.02 | 75.38 | |
| GPT-5Model Category=Closed-source models2026.06 | 75.3 | |
| DeepEyes-7BModel Category=Code-tool / visual-agent models2026.06 | 75.1 | |
| MiMo-VL-7B-RLnumber of parameters=7B2026.02 | 74.38 | |
| Pixel-Reasoner-7BModel Category=Code-tool / visual-agent models2026.06 | 74 | |
| MathCoder-VL-8BModel Category=Code-tool / visual-agent models2026.06 | 73.8 | |
| Qwen2.5-VL-32BModel Category=Open-source MLLMs (no visual tool), Instruct variant=true2026.06 | 73.4 | |
| Qwen2.5-VL-7Bnumber of parameters=7B2026.02 | 71.62 | |
| InternVL3-8BModel Category=Open-source MLLMs (no visual tool)2026.06 | 70 | |
| MiniCPM-V-4.5 (9B)number of parameters=9B2026.02 | 69.88 | |
| Qwen2.5-VL-7BModel Category=Open-source MLLMs (no visual tool), Instruct variant=true2026.06 | 68.8 | |
| GPT-5.12026.02 | 67 | |
| GPT-4oModel Category=Closed-source models2026.06 | 65 | |
| LLaVA-OneVision-7BModel Category=Open-source MLLMs (no visual tool)2026.06 | 63 |