The YC-backed lip sync platform built by the Wav2Lip team released sync-3, which generates a whole shot at once and re-syncs faces across many languages — with a free tier to try it.
2026-06-23 · by Moe Ameen
sync. (sync.so, from sync.labs) released sync-3, the latest and most capable model in its AI lip sync and visual dubbing lineup, in April 2026. The company positions sync-3 as a step up from its earlier models: instead of stitching together isolated frames, it builds an understanding of the speaker across the entire shot and generates the frames together, which is what lets it hold a convincing sync through close-ups, occlusions, extreme angles, and low light while preserving the original performance. It targets high-resolution output and works across a wide range of languages.
sync. is a focused platform: you upload a video, give it a voice — clone an existing one, generate a new one, or keep the clip's own audio — and the model re-renders the speaker's mouth to match. The same engine drives visual dubbing, where a translated voiceover gets matched to the on-screen presenter's lips so the result looks native rather than like an obvious overdub. The company came out of Y Combinator (W24), was founded by the team behind the open-source Wav2Lip project, and raised a seed round to expand the work. sync-3 sits alongside earlier models including the zero-shot lipsync-2, released earlier in 2025.
Access spans a web Lipsync Studio, an API with SDKs, and editor plugins for Adobe Premiere Pro and ComfyUI. There is a free way to try the platform, with paid plans that started around $5/month and climbed through higher tiers, each adding a per-second usage charge; watermark-free output begins on the paid Creator plan. As with any fast-moving model, treat specific resolution, language, and pricing figures as a snapshot and check sync.so for current details.
The fastest way to act on this is a localization play, and it pairs cleanly with Kompozy. Take your best-performing video, dub it in sync. into the languages your audience actually speaks — the presenter stays on screen, the mouth matches the new voiceover — then bring each language cut into Kompozy. Kompozy burns in captions in that language, reframes the clip for each platform, and schedules and publishes it across all nine destinations (TikTok, Reels, YouTube Shorts, Facebook, LinkedIn, X, Pinterest, Threads) per market, from one queue. One shoot becomes a multi-language, multi-platform rollout instead of a single upload.
Then go wider than the video. Because sync. only touches the mouth, the rest of each market's content is still on you — and that is exactly the gap Kompozy fills. Feed the same transcript in and Kompozy generates the supporting posts in your voice through a Persona Brief: a quote card, a text thread, a blog draft, a newsletter, a carousel — localized per market alongside the synced clip. sync-3 makes the dub look native; Kompozy turns each dub into a full week of on-brand, scheduled content across platforms.
sync-3 is sync.'s most advanced AI lip sync and visual dubbing model, released in April 2026. It builds a whole-shot understanding of the speaker and generates frames together rather than stitching them, which keeps the sync convincing through close-ups, occlusions, extreme angles, and low light.
sync. has a free way to try it, then paid plans that started around $5/month plus a per-second usage charge, rising through higher tiers. Watermark-free output begins on the paid Creator plan. Check sync.so/pricing for current numbers.
Yes. Feed a translated voiceover and sync-3 re-syncs the on-screen presenter's mouth to the new language, so the dub looks native instead of like an obvious overdub. It supports a wide range of languages; check sync.'s docs for the current list.
sync. returns a synced MP4 but does not publish it. Bring the export into Kompozy to add branded captions, reframe per platform, and schedule and publish across TikTok, Reels, YouTube Shorts, X, LinkedIn, and more — and fan the same clip into supporting posts in your voice.