Is Decagon Chord a public API? What Decagon has published
Decagon introduced Chord on October 1, 2026 as the speech model inside Voice 3. Its post does not state a public API, so a voiceover job needs another model.
Decagon has not stated a public API for Chord in the post I read: the article on Chord does not say whether it is a product release or an API, and it presents it as work from Decagon Labs. Decagon's Voice 3 post says Voice 3 launched October 1, 2026, with Chord as the speech model trained for customer conversations. If you need a voice for an ad or a short video today, plan around a model you can call.
Both Decagon pages were read on 2026-10-13. I am not citing outside write-ups for Chord's availability.
| Question | Answer from Decagon's pages |
|---|---|
| What is it? | A speech model post-trained for customer conversations |
| Architecture | Tokenizer-free diffusion |
| Launched with | Voice 3, October 1, 2026 |
| Public API stated? | Not stated in the Chord article |
| Languages (Voice 3) | 70+, with automatic switching mid-call |
What if I need a voice for a video this week?
Use the models that are in a catalog. Sume's TTS Router lists Cartesia Sonic rows (sonic-3.6, sonic-3.5, sonic-3, sonic-latest, sonic-preview) and ElevenLabs rows through fal (eleven-v4, eleven-v4-turbo, eleven-v3, eleven-multilingual-v2, eleven-turbo-v2.5). GET /v1/tts-router/models is the live list, and Chord is not in it.
How do I check any model before relying on it?
Ask the catalog, not a blog post. The router returns 400 with a catalog_url for an unknown model, so a typo fails fast and before any credit is reserved.
curl https://api.sume.com/v1/tts-router/models \
-H "Authorization: Bearer $SUME_API_KEY"Sources
Related posts
More in Models
- Eleven Multilingual v2 or Turbo v2.5 on Sume: 28 vs 31 languages
On Sume's TTS Router, Eleven Turbo v2.5 lists 31 languages, 20,000 characters and half the list price of Multilingual v2, which lists 28 languages.
- Every Sume image model compared: edit, text-only, n and formats
Sume's repo catalog lists 20 image rows: 14 take edit references, 5 are text-only, 1 makes layers. Grok is capped at n 1; Recraft is webp only.
- H3 Max lip sync and audio over 14.8 seconds: split or Fabric
Audio over 14.8 seconds is refused on Sume's H3 Max lip sync because the provider silently clips the rest. Fabric 1.0 takes the long line.
- H3 Max lip sync rejects audio under 5 seconds: what to do
Sume's MiniMax H3 Max lip sync takes 5 to 14.8 seconds of audio and returns invalid_request outside that window. Fabric is the fallback for shorter lines.
Written by Sume