Tenstorrent and Smallest.ai Launch On-Prem Production-Grade Voice AI Stack

Image: Tenstorrent · Source

Tenstorrent and Smallest.ai partnered to offer an on-prem voice AI system running Smallest.ai's Lightning V2 model on Tenstorrent Galaxy Blackhole servers, enabling cost-effective real-time voice applications with full data control.

Tenstorrent, an AI compute company, teamed with AI research lab Smallest.ai to deliver a production-grade, on-premises voice AI stack. This solution runs Smallest.ai's Lightning V2 real-time text-to-speech model natively on Tenstorrent Galaxy Blackhole servers, supporting enterprise-scale voice AI with high performance and reduced deployment costs. Deploying voice AI typically demands compromises among quality, cost, and data control. This partnership enables companies to run real-time voice agents fully on-prem, maintaining data sovereignty without sacrificing efficiency. According to Tenstorrent, its architecture offers up to four times efficiency gains at a fraction of comparable GPU costs, reshaping inference economics.