Scenario Open-Sources 1,300+ 3D Models and Generation Recipes
Scenario has released a 39 GB dataset of 1,300+ 3D models on Hugging Face under CC BY 4.0, featuring input images, prompts, 8K GLB, and Blender files.
AI 3D generation platform Scenario has open-sourced a comprehensive dataset containing over 1,300 3D models and their generation recipes, totaling 39 GB on Hugging Face under the CC BY 4.0 license. The collection features outputs generated across the Scenario platform using cutting-edge models including Hitem3D, Meshy, Rodin, Tripo, Claude Opus 5.5 3D, and GPT-6 Astra 3D.
Image source: Scenario (@Scenario_gg)
Dataset Composition and Specifications
A standout feature of this release is that every example goes beyond raw assets by shipping its exact generation metadata. Each entry includes the original input image, prompt, and precise parameter settings, enabling researchers and creators to audit and replicate generation workflows.
The dataset includes GLB files with textures up to 8K resolution, alongside native Blender files originating from Opus and Astra models. This rich structure gives developers and 3D artists immediate access to high-fidelity production assets for experimentation, fine-tuning, and project integration under open licensing.
Impact on the 3D AI Ecosystem
As multimodal 3D generation and AI agent pipelines mature, the industry requires robust open datasets to benchmark diverse models under uniform conditions. By releasing this 39 GB collection, Scenario provides the open-source community with an invaluable resource for analyzing the fidelity, texture quality, and structural integrity of modern 3D generators.
The dataset is available for direct download on Hugging Face.