Google launches Gemini 3.8 Live Avatar with real‑time lip‑sync in 97 languages for enterprises

On September 25, 2026, Google announced the launch of Gemini 3.8 Live Avatar, a synthetic presenter that can speak, lip‑sync and overlay on‑screen data in 97 languages without visual drift, and is now available to Gemini Enterprise customers.

The Live Avatar is positioned as an animated persona that reacts in real time, matching mouth movements and facial expressions to spoken output while also being able to display charts, bullet points or other visual cues directly on the screen.

Google stresses that the avatar can switch seamlessly among the 97 supported languages, and that language changes do not degrade video fidelity nor cause the visual drift that often plagues real‑time synthetic video.

How the technology works

Gemini 3.8 Live, introduced a week earlier, is a model capable of processing visual input near real time. The Live Avatar builds on that foundation by generating a three‑dimensional animated face that synchronises mouth shapes to the phonetic output of Gemini’s language model.

In the public demonstration, the avatar moved from English to Japanese mid‑sentence while keeping the mouth movements aligned with each language’s distinct phonemes, illustrating the system’s ability to handle language‑specific articulation on the fly.

The system also supports on‑screen overlays. While speaking, the avatar can display charts, bullet points or other visual cues, turning a simple voice response into a more complete presentation format.

Enterprise‑only rollout

Google has limited the Live Avatar to Gemini Enterprise customers at launch. Organizations that subscribe to the enterprise tier can either select from a catalogue of preset avatars or commission custom‑designed faces that match corporate branding guidelines.

Each piece of generated video carries an invisible SynthID watermark. Google says the watermark, together with built‑in safeguards, is intended to protect identity and help downstream platforms verify that the content is synthetic.

Potential benefits and use cases

• Customer support agents that can converse in the caller’s native language without hiring multilingual staff
• Internal training modules where a consistent presenter delivers material in multiple languages
• Marketing videos that can be localized instantly for global campaigns
• Live webinars where the avatar switches languages on the fly for multilingual audiences

The ability to maintain visual consistency across languages is particularly valuable for brand‑centric communications. Companies can avoid the uncanny‑valley effect that sometimes appears when subtitles are added to a static presenter, because the avatar’s mouth movements stay in sync with the spoken words.

However, the launch also raises identity, disclosure and accessibility questions. Regulators and advocacy groups have warned that synthetic presenters could be misused to impersonate real people or to obscure the source of information.

Google’s inclusion of a SynthID watermark is a first step, but the company has not detailed how the watermark will be detected by third‑party platforms, leaving an open gap in verification mechanisms.

From an accessibility standpoint, the avatar could improve content consumption for users with hearing impairments when paired with accurate captioning, while also offering visual cues for those who rely on lip‑reading.

Google has not yet announced specific accessibility certifications for the feature, so enterprises will need to evaluate compliance with regional accessibility standards on their own.

For an English‑speaking organization, the Live Avatar represents a scalable way to produce multilingual video content without the overhead of hiring voice actors, translators and video editors for each language.

A single Gemini Enterprise license can generate a synthetic presenter that delivers the same message in dozens of markets, preserving brand tone and visual style while cutting production time from weeks to minutes.

The result is faster go‑to‑market cycles, lower localization budgets, and a consistent viewer experience across the 97 supported languages, which could be a decisive advantage for global firms seeking rapid market entry.

In the Arab world, where multilingual communication often involves Arabic, French, English and several regional dialects, the Live Avatar could enable broadcasters and e‑learning platforms to produce a single video that speaks directly to each audience segment without re‑filming, thereby respecting cultural nuances while maintaining a unified visual identity.