AI 3D Model API: model comparison and examples

Generate textured 3D assets from text, a single image, or multi-view images through Muapi’s unified AI 3D Model API. This repository compares the live Tripo3D and Meshy endpoints and documents copy-paste HTTP examples for game assets, product visualization, AR/VR, and 3D-printing workflows.

Open the AI 3D Model API landing page · Get an API key · API reference

Related Projects

• Open Generative AI — curated open generative-AI tools and resources.
• Muapi ComfyUI — visual workflow integration for Muapi models.
• Generative Media Skills — reusable generative-media workflows.
• Awesome Generative AI Apps — application and workflow discovery.

What this repository covers

• Tripo3D H3.1, Tripo3D P1, and Meshy 6 text-to-3D, image-to-3D, and multi-image workflows.
• A comparison of input requirements, topology/material controls, output formats, and typical use cases.
• curl and Python examples for submitting an asynchronous job and polling its result.
• Request payload notes for prompts, image URLs, multi-view image lists, and output persistence.

Model comparison

Endpoint family	Inputs	Best for	Output notes
Tripo3D H3.1	Text, image, or 2–4 multi-view images	Strong price-to-quality and game-ready assets	Textured .glb; quality/topology options vary by endpoint
Tripo3D P1	Text or image	Premium Tripo3D generation	Textured .glb with premium quality tier
Meshy 6	Text, image, or 1–4 images	Configurable topology and export flexibility	.glb plus FBX, OBJ, and USDZ alternatives

Choose text-to-3D for concept exploration, image-to-3D for a single known reference, and multi-view generation when the back and sides of an asymmetric object matter. See the live page for current pricing and model availability.

Quick start with curl

export MUAPI_API_KEY="your_key"

curl -X POST "https://api.muapi.ai/api/v1/tripo3d-h31-text-to-3d" \
  -H "x-api-key: $MUAPI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"prompt":"a low-poly red panda astronaut, game-ready asset, clean studio lighting"}'

For image-to-3D, pass a publicly reachable image URL:

curl -X POST "https://api.muapi.ai/api/v1/meshy-6-image-to-3d" \
  -H "x-api-key: $MUAPI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"image_url":"https://example.com/product-front.jpg"}'

The submission response contains a request identifier. Poll until completion:

export REQUEST_ID="id_from_the_submission_response"
curl "https://api.muapi.ai/api/v1/predictions/$REQUEST_ID/result" \
  -H "x-api-key: $MUAPI_API_KEY"

See examples/quickstart.py for a complete polling loop.

API workflow

1. Create an API key and keep it server-side.
2. Select the endpoint that matches the input type and output requirements.
3. Submit POST /api/v1/{model-slug} with JSON input.
4. Poll GET /api/v1/predictions/{request_id}/result until the job succeeds or fails.
5. Copy the output URL into durable storage before a temporary result expires.

For multi-view reconstruction, use the endpoint’s documented images_list input and keep the images consistently framed and lit. Do not commit reference images, generated assets, or API keys to this repository.

Common failure modes

Symptom	Check
Validation error	Endpoint slug, required image field, reachable URL, and JSON array shape.
Inconsistent geometry	Use 2–4 distinct views with a shared subject, scale, and background where possible.
Missing export format	Meshy provides the broadest documented export set; inspect the result payload before assuming every format is present.
Unknown job state	Poll the result endpoint instead of submitting duplicates; persist the request ID.

License

Documentation and examples are MIT licensed. Generated meshes, textures, and model-provider usage remain subject to the applicable Muapi and provider terms.