Build with ManifoldGen

Quickstart

Send your key as Authorization: Bearer …. Keep it on your server; never ship it in public browser code.

Generate video

Creates an asynchronous video job. Audio is enabled by default.

curl https://manifoldgen.com/api/service \
  -X POST \
  -H "Authorization: Bearer $MANIFOLDGEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "service": "video",
    "prompt": "A glass greenhouse at night, rain on the roof",
    "aspect_ratio": "16:9",
    "size": "native",
    "duration": 5,
    "num_steps": 20,
    "include_audio": true,
    "output_format": "webm-av1"
  }'

202 Accepted

{
  "result": {
    "job_id": "8a17…",
    "status": "queued",
    "status_url": "/api/video-jobs/8a17…"
  },
  "estimated_credits": 101,
  "estimated_cost_usd": 1.01
}

Poll a job

Poll every 2–5 seconds until status is completed. Completed responses include the durable video URL, actual cost, and credits used.

curl https://manifoldgen.com/api/video-jobs/8a17… \
  -H "Authorization: Bearer $MANIFOLDGEN_API_KEY"

Generate images

Image requests return synchronously and start at 4 credits.

curl https://manifoldgen.com/api/service \
  -X POST \
  -H "Authorization: Bearer $MANIFOLDGEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "service": "image",
    "prompt": "Editorial portrait, hard side light",
    "width": 1024,
    "height": 1024,
    "num_steps": 12
  }'

Generate Anima character art

One anime-native illustration per asynchronous request for a fixed $0.04. Check GET /api/anima/status before dispatch; the route fails closed until commercial model licensing and GPU capacity are both active.

curl https://manifoldgen.com/api/service \
  -X POST \
  -H "Authorization: Bearer $MANIFOLDGEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "service": "anima",
    "prompt": "An adult celestial cartographer, midnight-blue hair, brass astrolabe",
    "negative_prompt": "text, watermark, low detail",
    "width": 768,
    "height": 1024,
    "num_steps": 28,
    "guidance": 4,
    "seed": 18467291
  }'

Canvas sides are 512–1536 pixels, divisible by 16, with at most 1,048,576 total pixels. Poll the returned video-job status URL; the completed result contains image_url.

Generate music and sound effects

Use the audio umbrella with a kind, or call the dedicated music and sfx services. Both persist as durable audio assets and become searchable by meaning.

Audio umbrella

curl https://manifoldgen.com/api/service \
  -X POST \
  -H "Authorization: Bearer $MANIFOLDGEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "service": "audio",
    "kind": "music",
    "prompt": "Warm modular synths, restrained drums, hopeful sunrise",
    "duration": 30
  }'

Music · synchronous · 30–300 seconds

curl https://manifoldgen.com/api/service \
  -X POST \
  -H "Authorization: Bearer $MANIFOLDGEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"service":"music","prompt":"Slow cinematic strings at dawn","lyrics":"[Verse]\nNeon rain over the harbour","duration":45,"service_tier":"standard"}'

200 OK

{
  "service": "audio",
  "audio_id": "9c41…",
  "audio_url": "https://…/track.wav",
  "kind": "music",
  "duration_seconds": 30,
  "indexed": true,
  "credits_used": 80,
  "cost_usd": 0.80
}

SFX · asynchronous · 4–45 seconds

curl https://manifoldgen.com/api/service \
  -X POST \
  -H "Authorization: Bearer $MANIFOLDGEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"service":"sfx","prompt":"Heavy wooden door closing in a stone hall","duration":5}'

202 Accepted

{
  "service": "sfx",
  "result": {
    "job_id": "video_71c9…",
    "status": "queued",
    "status_url": "/api/audio-jobs/video_71c9…"
  },
  "estimated_credits": 51,
  "estimated_cost_usd": 0.51
}

Poll the returned status URL. A completed SFX job contains audio_id, audio_url, final cost, and credits used.

Search public audio

Search needs no key for public assets. Add your Bearer key to include your private audio in the results.

curl --get https://manifoldgen.com/api/audio/search \
  --data-urlencode "q=hopeful modular sunrise" \
  --data-urlencode "kind=music" \
  --data-urlencode "top_k=20"

Generate voices

Generate one to four durable voice takes with Seed Audio, ElevenLabs, Qwen, MiniMax, Seed Speech, Google, or Grok. Call GET /api/voice/models for live pricing and model-specific controls.

curl https://manifoldgen.com/api/voice/generate   -X POST   -H "Authorization: Bearer $MANIFOLDGEN_API_KEY"   -H "Content-Type: application/json"   -d '{
    "model": "seed-speech",
    "text": "The future sounds closer than you think.",
    "voice": "stokie_en",
    "voice_details": "Warm, confident, cinematic delivery",
    "mood": "happy",
    "batch_size": 1,
    "speed": 1,
    "pitch": 0,
    "volume": 1,
    "output_format": "mp3",
    "sample_rate": 24000
  }'

Clear, usage-based pricing

Choose a quality tier and duration up front. Every request returns the same preflight dollar and credit estimate shown below; completed video jobs settle from measured generation time and include the final charge.

Errors