01 / AI · Analysis
Bland unveils Tofu-1 to listen while speaking on calls
A DEV.to report described simultaneous listening and speaking on calls and Bland's plans for a family of speech-to-speech models.

Bland unveiled Tofu-1 on 13 Nov as a voice model for high-stakes calls, according to a DEV.to report.
The report described a model that listens to the other caller while delivering its own response. That makes simultaneous listening and speaking the central feature of the call interaction it outlined.
According to DEV.to, Bland co-founder and CEO Isaiah Granet announced the release on X. The report also presented Tofu-1 as the starting point for a planned family of speech-to-speech models.
Listening while speaking
DEV.to described Tofu-1 as a full-duplex voice model, meaning it handles listening and speaking at the same time. In the interaction described, listening continues while the model delivers a spoken answer.
The report also attributed reasoning during speech to Tofu-1. Its account therefore covered both receiving the caller's audio and continuing to reason while a response is being spoken. These were reported capabilities, rather than independently confirmed findings.
A planned family of voice models
According to DEV.to, Bland planned a family of speech-to-speech models beginning with Tofu-1. Speech-to-speech describes systems that take spoken audio as input and produce spoken audio as output.
The report positioned Tofu-1's combination of listening, speaking and reasoning during calls as the starting point for that family. The release concerned Tofu-1; the broader model family remained a plan.