Blog

Dialt now offers on-prem voice assistants

· Ronan McGovern

Dialt now offers self-hosting on premises or in your own cloud, on a single GPU or across multiple GPUs. We work with your team on dialects, custom voices and conversational behaviors, with the deployment scoped to your application.

Our offerings can fit on a single H100, or B200 for increased concurrency and reduced latency.

Latency is typically lower than our hosted solution, while capabilities are close, and competitive with leading hosted realtime voice providers.

Eight historical voice benchmark results: Dialt Fast, Smart, On-Prem H100 and B200, GPT Live-1, Gemini 3.8 Live, Eleven Realtime Mini and Max. Dots show p50 response time, horizontal whiskers p90, and vertical bars 95% score bootstrap intervals. Faster is to the right. Recording clocks and cohorts differ.

Larger chart · Chart data and measurement notes

To discuss self-hosting Dialt on premises or in your own cloud, get in touch at hello@dialt.com.