The Qwen Team has released BlenderWorld on ModelScope, a benchmark testing whether visual agents can turn real-world references into editable 3D scenes. It contains 223 Blender tasks across 9 categories, with text, image, and video inputs. The benchmark includes 2,651 task-specific evaluation rubrics.
🌍 Meet BlenderWorld from the Qwen Team, now on ModelScope!
Explore 👉 https://modelscope.ai/datasets/Qwen/BlenderWorld
Can visual agents turn real-world references into editable 3D scenes?
🧊 223 Blender tasks across 9 categories
🖼 Text, image & video inputs
📏 2,651 task-specific evaluation rubrics
Source: ModelScope · x.comPublished
