Introducing Interactive Avatars that can listen, talk, and respond directly inside a product, website or app: We’re taking an open approach: you can combine your own LLM + voice stack with @synthesiaIO's API to create an integrated system that can be used as a support concierge, a live FAQ agent, or anywhere else someone needs a conversational interface that actually listens and responds. The avatars are multilingual and support idle and listening behaviour, so conversations flow smoothly even during pauses. You can also build completely customizable avatars from photorealistic photos or branded characters like an animated mascot, with adjustable backgrounds. Finally, our avatars support agentic calls like scheduling a meeting or filling a form at the end of the conversation so you can convert conversations into real business opportunities.

Today, we are introducing Express-3, a new video model that delivers sharper, more expressive Avatars while also cutting video creation time in half. In addition to our classic photorealistic Avatars, Express-3 supports a new category of Avatars we're calling Style Avatars: stylized, cartoonish characters that can be created from a prompt. Style Avatars have been a long requested feature (a lot of companies have their own mascots!) and are great for creating more accessible content. Express-3 Avatars work seamlessly with @synthesiaIO Assistant. Generate an Avatar, then include it in a video and see how the Assistant instantly matches the visuals in the video to the Avatar's characteristics!