Embodied Planning & Correction on Cosmos
69.3AccuracyEmbodied-R1.5
Evaluation Results
| Method | Links | |
|---|---|---|
| Embodied-R1.5Model Type=Ours2026.06 | 69.3 | |
| GPT-5.4Model Type=General-Purpose VLM2026.06 | 69.2 | |
| Qwen3-VL-8BModel Type=General-Purpose VLM2026.06 | 64 | |
| Gemini-2.5-ProModel Type=General-Purpose VLM2026.06 | 62.8 | |
| Pelican-VL-1.0Model Type=Embodied VLM2026.06 | 62.6 | |
| VeBrainModel Type=Embodied VLM2026.06 | 58.8 | |
| Mimo-EmbodiedModel Type=Embodied VLM2026.06 | 56.8 | |
| Embodied-R1Model Type=Embodied VLM2026.06 | 55.2 | |
| Gemini-Robotics-ER-1.5Model Type=Embodied VLM2026.06 | 53.7 | |
| GPT-4oModel Type=General-Purpose VLM2026.06 | 53.3 | |
| InternVL3.5-8BModel Type=General-Purpose VLM2026.06 | 48.2 | |
| RoboBrain-2.0-7BModel Type=Embodied VLM2026.06 | 33.8 | |
| MagmaModel Type=Embodied VLM2026.06 | 32.8 | |
| Molmo-D-7BModel Type=General-Purpose VLM2026.06 | 22.2 |