Cartesia Sonic 3.6 API: which model id to send
Send sonic-3.6 for the latest stable Sonic 3.6, a dated id to freeze it, or sonic-preview for beta. Which of these ids Sume's TTS Router lists.

For the Cartesia Sonic 3.6 API, send the model id sonic-3.6 to follow its most recent stable snapshot, a dated id like sonic-3.6-2026-08-27 to freeze the behavior, or sonic-preview to test upcoming changes. Cartesia's docs say the dated snapshot never changes and that sonic-preview is a beta model, not for production.
The Cartesia facts are from its Sonic 3.6 docs page, read 2026-09-29. The Sume facts are from the TTS Router in the Sume API reference. Sume's router lists five Sonic ids and no dated ones.
What does each Sonic 3.6 model id do?
Cartesia calls Sonic 3.6 generally available and dates the first stable snapshot August 27, 2026. Its docs give three kinds of id.
| Model id | Update behavior (Cartesia) | Recommended for |
|---|---|---|
sonic-3.6-YYYY-MM-DD | Snapshotted, will never change | Customers who want to run internal evals before any updates |
sonic-3.6 | Will be updated to point to the most recent stable snapshot | Customers who want stable releases and recent capabilities |
sonic-preview | Beta model, will always be updated with upcoming changes | Testing purposes |
Which of these ids does Sume's TTS Router accept?
POST /v1/tts-router/generate requires a model from GET /v1/tts-router/models. The model field is an enum of five ids. A dated sonic-3.6-YYYY-MM-DD id is not one of them, so a request with it does not match the schema.
Sume's TTS 1.0 route has no engine picker: it rejects model with a 400. Use the router when you want to name the Sonic version.
| Sume `model` id | Note in the catalog |
|---|---|
sonic-3.6 | Sonic 3.6 |
sonic-3.5 | Sonic 3.5 |
sonic-3 | Sonic 3 |
sonic-latest | Alias for sonic-3.6, never sonic-preview, and not a dated snapshot |
sonic-preview | Beta channel: output and availability can change without notice. Not compatible with pro voice clones (voice_model_mismatch) |
How do I call Sonic 3.6 through Sume?
Send model, a transcript, a voice and, for a non-English script, a language. The voice is a ready avatar's (avatar_id or avatar_handle) or a voice.id. The response is a job; the audio is a Sume-hosted file when it completes. Text to speech API covers voice selection and the rest of the body.
curl -X POST https://api.sume.com/v1/tts-router/generate \
-H "Authorization: Bearer $SUME_API_KEY" \
-H "Content-Type: application/json" \
-H "Idempotency-Key: sonic-36-001" \
-d '{
"model": "sonic-3.6",
"transcript": "Your order ships tomorrow.",
"avatar_handle": "acme",
"language": "en"
}'What are the limits and the price?
Every catalog row lists a max_characters capability, and the request schema caps transcript at 20,000 characters, spaces and punctuation included. The router charges on the same character book as TTS 1.0. Today that is $0.0475 per 1,000 characters plus a 5.5% agent fee by default; see API pricing.
- Pick
sonic-3.6orsonic-latestfor the current stable release;sonic-latestis an alias, so both resolve to the same id today. - Do not use
sonic-previewfor production audio: its output can change without notice. - Audio that runs past 1,200 seconds fails with
tts_duration_exceeded. Split long scripts. - Unknown ids fail with
400 model_not_foundand the catalog URL.
Sources
Related posts
More in Models
- AI outfit change video: change clothes in a clip with AI
Change someone's outfit in an existing video with an AI video-to-video edit: name the garment, what it becomes, and what must stay the same.
- ChatGPT Images 2.5 vs GPT-Image-2.5: which name is the API?
ChatGPT Images 2.5 is OpenAI's product; GPT-Image-2.5 Flare and Sunburst are the API models. How the names map to Sume ids and which features stay in ChatGPT.
- Does ChatGPT Images 2.5 add a watermark? C2PA explained
OpenAI says ChatGPT Images 2.5 uses C2PA metadata and invisible watermarking to identify its images. What that means for files you get through an API.
- Claude Sonnet 5.5 or Opus 5.5 for an agent that calls a video tool?
Sonnet 5.5 and Opus 5.5 differ in price and release date on Anthropic's pages. Which facts matter when the tool is Sume's video generation, and what stays same?
Written by Sume