product-launch · August 3, 2026
By Joe Jurczyk, Editor
UNITH announces alpha release of DEVA-1 avatar model, beta set for September
ASX-listed UNITH disclosed an alpha release of DEVA-1, a generative model for digital human avatars supporting single-image and video avatar generation with speech-tone-driven real-time streaming. Beta access for paying clients was scheduled for September 2026.
UNITH Ltd (ASX:UNT) announced the alpha release of DEVA-1 on 3 August 2026, describing it as a new generative AI model for digital human avatars.
According to the announcement as reported, DEVA-1 supports avatar generation from a single image or from video, real-time streaming driven by speech tone, and reduced conversational latency, with 3D rendering that responds to speech characteristics. UNITH said a beta giving production-ready access to paying clients was scheduled for September 2026, with full rollout expected by the end of 2026. The company said eight alpha partners characterised DEVA-1 as a "genuine generational leap" and that several indicated interest in migrating existing deployments. UNITH cited a digital humans market of US$7–10 billion.
Why it matters for digital humans: Single-image avatar creation and prosody-driven animation are now table stakes claims across this category, so the interesting question is not the feature list but whether the latency holds under load — and no figures were published. Every substantive claim here is company-reported in a listed-company disclosure, including the alpha-partner quotes and the market size, and none has been independently verified. The September beta date is the checkable milestone: if paying clients are on it, that is evidence; the alpha testimonials are not.