3DHarnessBench: Probing Agentic 3D-to-Code Capabilities of Frontier Vision-Language Models
We introduce 3DHarnessBench, a benchmark that evaluates the agentic ability of frontier vision-language models (VLMs) to recover 3D geometry as Blender Python code from a variety of inputs. Unlike previous frameworks that prompt the VLMs with a fixed input (e.g., a single rendering or a text description), 3DHarnessBench evaluates four separate harness settings that progressively enable active agentic exploration, facilitated by recent Blender MCP functionality. Our hierarchy from Single-view, Multi-view, Active Visual (arbitrary viewpoint access), and Full 3D Interaction (complete access to the target object through Blender function calls) probes the models' abilities in both visual perce...
| Authors | Ling Liu, Bingchen Gong, Amal Dev Parakkat et al. |
|---|---|
| Primary category | cs.CV |
| Published | 6 September 2026 |
| Last updated at source | 6 September 2026 |
Connected in the knowledge graph
- lingada/3DHarnessBenchDataset · cited by
Models and datasets referencing this paper
- lingada/3DHarnessBenchDataset
Silytics shows public metadata from arXiv and links to the original. Information may be incomplete or out of date. Last refreshed 17 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.