Tofu-1
Bland released Tofu-1, a full-duplex speech-to-speech model for phone calls, in a closed enterprise research preview.
Most voice agents chain speech to text, a language model and text to speech. Tofu-1 takes audio in and gives audio out with no transcript between, so tone, hesitation and background sound stay available to the model. It can be interrupted mid-sentence and keeps the conversation going while it works on a request. Bland has not published the architecture, size, latency or benchmarks.
- Date
- Sunday 15 November 2026
- Lab
- Bland
- Kind
- model
- Access
- research preview
Figures
| Measure | Value | Measured by |
|---|---|---|
| Screenings where listeners could not spot it as a model | over 92% self-reported, as relayed by MarkTechPost | company |
Closed research preview for a limited number of enterprises, with Bland citing January availability. Only secondary coverage was read.
Sources
This record was checked against its sources on 15 November 2026. How we check