ModelComputer Visionvv1

Shap-E

by OpenAI · open-source · Last verified 2026-03-17

Shap-E is OpenAI's improved 3D generation model that generates implicit neural representations of 3D objects (which can be rendered as textured meshes or NeRFs) from text or image prompts, producing richer geometry and appearance than its predecessor Point-E. It encodes 3D assets into a compact latent space and trains a diffusion model over that space, enabling more coherent shape-and-texture output.

https://github.com/openai/shap-e ↗

C+

C+—Average

Adoption: C+Quality: B+Freshness: C+Citations: B+Engagement: F

Specifications

License: MIT
Pricing: open-source
Capabilities: text-to-3d, image-to-3d, nerf-generation, textured-mesh-export, implicit-neural-representation
Integrations: huggingface, pytorch
Use Cases: 3d-asset-prototyping, game-development-research, product-visualization, research, creative-exploration
API Available: Yes
Parameters: ~300M
Context Window: N/A
Modalities: text, image, 3d
Training Cutoff: 2023
Tags: 3d-generation, text-to-3d, implicit-representation, openai, open-source
Added: 2026-03-17
Completeness: 100%

Index Score

52.8

Adoption

Quality

Freshness

Citations

Engagement

Explore the full AI ecosystem on Agents as a Service