Mcandymuse

Production API · Controlled beta

Prompt in.
Creation out.

Create images from a required text prompt, optionally guided by a reference. Generate native-audio video from a reference image and a supported preset.

Read the docs Request API access
  • Asynchronous tasks
  • Idempotency-Key retries
  • Private media
Create an image from a prompt
curl -X POST https://candymuse.ai/api/create_image \
  -H "API-KEY: $API_KEY" \
  -H "Idempotency-Key: your-unique-request-id" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "A cinematic night portrait with soft neon light",
    "quality": "low"
  }'
HTTP 200status: NEW
Image inputPromptreference image optional
Image quality$0.020 / $0.025low / high per accepted task
Native A/V$0.1505-second low-quality request
ExecutionAsyncdedicated GPU workers

Two production pipelines

Choose the output. Keep one task model.

Both routes share authentication, idempotency, polling, billing, and private result delivery.

Image API

Create from text. Guide with a reference.

Required prompt · optional reference image · low or high quality
  • prompt-first generation
  • optional reference guidance
  • immutable result receipts
View image endpoint →

Video API

Generate motion with sound.

Six verified compatibility presets · H.264 + native AAC
  • 5-second output
  • six compatibility presets
  • durable asynchronous execution
View video endpoint →

A small integration surface

Create. Poll. Download.

  1. 01

    Describe the result

    Send a required image prompt and optionally include a PNG, JPEG, or WebP reference. Video requests use a reference and preset.

  2. 02

    Poll the task

    Creation is asynchronous. Poll one durable task ID until it completes.

  3. 03

    Stream the result

    Download through a short-lived CandyMuse URL with full and Range support.

Ready to integrate?

Start with one curl command.

The current invite-only beta supports a deliberately narrow, verified production contract.

Open developer docsContact CandyMuse