Veo 3 adds Gemini 3.8 Live Avatar

Veo 3 adds Gemini 3.8 Live Avatar, letting creators feed a text prompt and receive a lifelike talking‑head video that syncs speech to facial motion in real time. Google Veo 3, the AI video‑generation platform, expands its toolkit with this live‑avatar capability, closing a gap for users who previously relied on static renders or third‑party avatar services.

What's new

Google Veo 3 now includes the Gemini 3.8 Live model, which generates a full‑body or head‑only avatar that lip‑syncs to input audio with sub‑second latency. The feature supports resolutions up to 1080p at 30 fps and can run on a single V100 GPU, making it accessible to cloud‑based creators. As of September 2026, the live‑avatar option rolls out to all Veo 3 Pro subscribers and is available in a limited beta for free‑tier accounts.

How it fits your workflow

Filmmakers can replace costly on‑set talent with a Veo 3‑generated avatar that delivers consistent performance across multiple takes. Editors gain the ability to swap dialogue on the fly without re‑shooting, streamlining post‑production. Compared to tools like Synthesia 2 and Runway Gen‑4, Veo 3’s live avatar runs directly in the Veo interface, eliminating the need for separate export‑import steps. Streamers and educators also benefit, as they can create personalized on‑camera presenters without a physical studio, reducing production overhead while maintaining a human‑like presence.

What it costs / how to try it

Google Veo 3’s Gemini 3.8 Live Avatar is included in the existing Pro plan at $49 per month; free‑tier users can request beta access via the Veo dashboard. No additional licensing fees apply for the live‑avatar feature.