Frontier3DBench benchmarks coding agents for 3D creation
Original title: Frontier3DBench: Benchmarking Coding Agents for 3D Creation across Objects, Scenes, and Interactive Open Worlds
GPT-6.1-sol leads object and scene levels on Frontier3DBench while reference reconstruction remains a key bottleneck
Frontier3DBench evaluates 11 model-harness configurations across 145 tasks in object, scene, and interactive world creation. GPT-6.1-sol achieves the highest overall score of 66.02/100 on object and scene tasks. All configurations score lower on reference reconstruction than text-driven generation, and coordinated activity completion challenges persist.
- 145 tasks across 9 categories
- GPT-6.1-sol scores 66.02/100 overall
- GPT-6-astra leads interactive world construction
- Reference fidelity and motion clearance are bottlenecks
From the original article
1 min read at huggingface.co
Coding agents combine visual understanding with program execution, offering a new route to automated 3D creation. By translating images and instructions into editable programs, they bring reconstruction, generation, and editing into a shared workflow. Yet existing benchmarks fragment evaluation across task scopes and protocols, leaving their capabilities across construction levels unclear.
By Xinjie Lin, Qingle Liu, Wenyi Li · huggingface.co
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.