# Is Fish Audio Live 3 SOTA?

Pending benchmarks. Announced Nov 10, 2026; no independent results yet.

As of Nov 22, 2026, Fish Audio Live 3 is pending benchmarks on isitsota: announced Nov 10, 2026, with no independent results on the boards we track yet.

Fish Audio's full-duplex voice model for real-time voice agents. It keeps listening while it speaks, so a pause, a short acknowledgment, or a change of mind can shape what happens next. Announced November 10, 2026. It speaks English today, and is available on the Fish Agent Platform and through the Fish Audio API; enterprises can deploy it on-premises. Open weights were not stated, and no public weight release was found.

Canonical: https://isitsota.com/m/fish-audio-live-3
Last updated: 2026-11-10
Tracking since: 2026-11-10

## Announced facts

| Fact | Value |
| --- | --- |
| Lab | Fish Audio |
| Announced | Nov 10, 2026 |
| API | Live |
| Open weights | Unknown · Not stated; no public Hugging Face or GitHub weight release found |
| Type | Full-duplex voice model for real-time voice agents; trained on real human conversations so it learns when to acknowledge, wait, respond, or stop (Fish Audio blog) |
| Languages | Speaks English today (Fish Audio blog) |
| Hosted access | Available on the Fish Agent Platform and through the Fish Audio API; try at fish.audio/agents (Fish Audio blog). Fish Agents is in public beta and every Fish Audio account can use it (Fish Audio docs). Docs do not yet list a dedicated Live 3 model ID as of this entry |
| On-premises | Enterprises can deploy it on-premises (Fish Audio blog) |
| License | Not stated |
| Related | Separate from Drama 3 and Fish Audio Realtime STT (transcribe-stream-1) |

Tags marked unsure are not stated on the lab's own pages.

## What would let us call it

Verdicts use only independent boards we already track. Two consecutive snapshots are required before any crown.

- [x] Publicly announced with a source
- [x] Usable by the public (API or weights)
- [ ] Listed on an independent board we track
- [ ] Enough results for a rank in a category
- [ ] Holds across two snapshots (crowns only)

## What the lab claims

Self-reported. Shown for context; not used in any verdict.

- Lab claim: can decide to respond in as little as 186 milliseconds (Fish Audio blog).

## Sources

- LAB: [Fish Audio blog: Introducing Live 3](https://fish.audio/blog/introducing-live-3/) (2026-11-10)
- LAB: [Fish Audio docs: Agents Overview](https://docs.fish.audio/agents/overview)

Tracking continues until an independent board lists it.

## Related

- Lab: [Fish Audio](https://isitsota.com/labs/fish-audio.md)
- All pending releases: https://isitsota.com/releases.md
- JSON: https://api.isitsota.com/v1/models/fish-audio-live-3
