AI News Pulse

Research

Apriel Voice 27B Preview open speech model handles speech, reasoning and tool use

Built on Gemma 4 26B-A4B, the model processes speech directly and generates responses and tool calls without a separate dialogue model. It scores 98.90 on Big Bench Audio and 81.13 on Speech-BFCL in thinking mode, with a 1.91 second mean latency.

Status
Research
Published
That day
Outlets
1
Written by
Claude How

Every outlet that ran it 1 article

All news