pinwing.ai Pricing Sign in Start free

A rendering API for 3D models

POST a GLB and get back a turntable video, rendered on GPU nodes and delivered as files plus a share link. One call covers the simple case; batch and safe retries cover pipelines. 

# upload a model + render it, one call
curl -X POST https://api.pinwing.ai/v1/models \
  -H "Authorization: Bearer pinwing_sk_…" \
  -F file[email protected] \
  -F renderSettings='{"quality":"share","length":6}'

Omit renderSettings and the upload's render is a free 540p preview instead.

Built for unattended pipelines

  • •

    Safe retries. Idempotency-Key replay on render creation: a timed-out submit retried with the same key returns the original render, uncharged.

  • •

    Batch. Submit many aspect/background/duration variants in one atomic call; poll the whole batch with delta queries.

  • •

    Quote before you spend. A quote endpoint prices any settings object before you commit credits; cancellation refunds unstarted work.

  • •

    Everything downloadable. Video, web variant, poster, and thumbnail, streamed or via presigned direct-from-storage URLs; the zip bundle streams on the fly.

Rendered through the API

Mech Drone9s
Terrarium Bots6s
Sony FX-300 Jackal6s

Render API — common questions

Bearer API keys with the pinwing_sk_ prefix, created in the studio app (legacy dxgl_sk_ keys stay valid). Keys carry scopes and can be personal or bound to a team workspace. One header on every request, with no OAuth dance.
Yes: a full OpenAPI 3.1 description of every endpoint at pinwing.ai/openapi.json, plus an RFC 9727 API catalog at /.well-known/api-catalog. Point your codegen or agent tooling at either.
Yes. The hosted MCP server at mcp.pinwing.ai exposes the pipeline as tools for Claude Desktop, Cursor, Windsurf, and other MCP clients — import models, quote costs, render, batch, and download from natural-language prompts, using the same API keys.
Send an Idempotency-Key header on render creation: a retry with the same key and identical settings replays the original response instead of creating (and charging) a second render. Batch submissions accept the same header.

The full reference covers every endpoint, setting, and error code.

Read the API docs MCP server Get an API key