Sync.so
Lip sync, dubbing and image-to-talking-video for footage you already shot
sync. re-syncs a speaker's mouth to new audio in existing footage, and now also translates and dubs a video with cloned voices in one pass. The sync-3 model (April 2026) outputs up to 4K at 60fps and is available through a web studio, a REST API with SDKs, an MCP server and a Premiere/DaVinci Resolve plugin pair. Paid plans bill a subscription plus per-second usage.
- 2 creators using
- 2023 first launched
- Freemium pricing
- Easy learning curve
What Sync.so can do
- Infinite canvas No
- AI agent No
- Node / flow builder No
- Public API Yes
- MCP server Yes
11 of 26 capabilities confirmed
| Capability | Supported | Details |
|---|---|---|
| Text to image | No | |
| Image editing | No | |
| Text to video | No | |
| Image to video | Yes | Image-to-video on sync-3 (June 2026): one still plus audio becomes a talking video. |
| Video to video | Yes | Lipsync on existing footage is the core product; dubbing re-syncs translated speech. |
| Native audio in video | No | Syncs mouth to supplied audio; does not generate the soundtrack in the same pass. |
| Lip-sync | Yes | sync-3, lipsync-2-pro, lipsync-2, react-1 and lipsync-1.9.0-beta. |
| Voice / TTS | Yes | Voice cloning and text-to-speech for dubbing (own TTS key on Creator and above; 10 free TTS generations per month). |
| Music & SFX | No | |
| Upscaling | No | |
| Character consistency | Partial | Zero-shot: keeps the on-screen speaker's performance and style without per-person training, but there are no persistent character references across generations. |
| Camera control | No | |
| Custom training | No | Zero-shot, so no per-speaker training. |
| Third-party models | Partial | Dubbing uses ElevenLabs voices; the lipsync models are sync's own. |
| Capability | Supported | Details |
|---|---|---|
| Infinite canvas | No | |
| Node / flow builder | No | Has a multi-step dashboard and a timeline editor, but no node or flow builder. The ComfyUI node is a plugin into ComfyUI, not a sync. builder. |
| AI agent | No | Assistant mode picks a model and detects speakers from a request, but it runs one lipsync job rather than planning multi-step work. |
| Chat assistant | Partial | sync. assistant (beta) guides a single lipsync; Assistant mode in the studio; ChatGPT plugin. |
| Team collaboration | Partial | Team workspaces on Growth and up, 3 seats included on Growth, extra seats $25/month. |
| Capability | Supported | Details |
|---|---|---|
| Public API | Yes | REST API at api.sync.so/v2 (x-api-key), official Python and TypeScript SDKs, webhooks, Batch API on Scale, estimate-cost endpoint. |
| MCP server | Yes | Official open-source MCP server @sync.so/mcp-server; works with Claude, ChatGPT, Cursor, Windsurf and Codex. |
| Mobile app | Not confirmed | |
| Desktop / local | No | Cloud service. Local use is only through the DaVinci Resolve and Premiere plugins, which call the cloud API. |
| Open source | No | The MCP server is open source; the models are not. |
| Capability | Supported | Details |
|---|---|---|
| Free tier | Yes | 3 generations per month, 20-second max per clip, watermarked output |
| Commercial use | Not confirmed |
Yes Partial No Not confirmed by the vendor
What makes Sync.so different
- 01
Lipsync is the whole product: it works on footage you already shot, so you don't regenerate the performance or the shot.
- 02
sync-3 is stated to output up to 4K at 60fps, and the model handles multiple speakers by identifying who is talking before syncing the face.
- 03
Official MCP server (@sync.so/mcp-server), Python and TypeScript SDKs, and a REST API at api.sync.so/v2 with a Batch API on the Scale plan.
- 04
Native hooks into editors: a DaVinci Resolve Studio plugin, an Adobe Premiere plugin and a ComfyUI node, so the lipsync can run without leaving the timeline.
- 05
Per-second pricing with no credits to pre-buy: the model rate is the same on every plan, and plans only change the discount (up to 20% on Scale) and the max clip length.
About Sync.so
sync. is a lipsync model and workflow product rather than a generator: you bring a video and an audio track (or a script and voice) and it returns the same shot with the mouth matched to the new speech. It is used for fixing a line in a take, matching a new voiceover, redubbing for other languages, and turning a single photo into a talking clip.
What changed in 2026: sync-3 launched on 6 April and is the default model, with the claim that it reads the whole shot at once (face position, lighting, head angle, who is speaking) rather than stitching small chunks. In May the company added Dubbing (translation, voice cloning and lipsync in one step, 29 languages) and an Assistant mode that picks a model and detects speakers from a plain-English request. June brought image-to-video on sync-3 and a DaVinci Resolve Studio plugin. July added a ChatGPT plugin, and on 24 July the dashboard was rebuilt around creation tasks with a timeline editor for multi-clip lipsync and dubbing.
For a film or video workflow it sits after the shoot and after generation: fix a performance, redub a spot for three markets, or lip sync an AI-generated character to a recorded line. It is not a text-to-video tool and has no canvas or node builder.
What's new in Sync.so
-
Dashboard rebuilt around creation tasks, with a timeline editor
The new dashboard offers lipsync, translate-and-dub, image-to-video and a multi-track timeline editor, and lets one project chain several edits.
Source: sync.so ↗ -
sync. is available inside ChatGPT
A ChatGPT plugin lets users lipsync, dub and clone voices from a prompt.
Source: sync.so ↗ -
Lipsync inside DaVinci Resolve Studio
A plugin picks clips from the Resolve media pool, runs a sync. model and returns the result to the edit.
Source: sync.so ↗ -
Image-to-video on sync-3
Upload a single still plus audio and get a lip-synced talking video with no source footage.
Source: sync.so ↗ -
AI dubbing and Assistant mode
Translate, clone the voice and lipsync in one workflow across 29 languages; Assistant mode chooses the model and speakers from a plain-English request. MCP server and sync-3 for free users were announced in the same changelog entry.
Source: sync.so ↗ -
sync-3 launch
New flagship lipsync model, up to 4K at 60fps, aimed at close-ups, low light, sharp angles and multi-speaker shots.
Source: sync.so ↗
Models available in Sync.so
-
lipsync-2 — Sync Labs Lowest-cost lipsync model; fast Also: lipsync-2-pro · lipsync-1.9.0-beta
-
-
-
ElevenLabs voices $$$ ElevenLabs Voice source used by the Dubbing workflow
$ · $$ · $$$ = relative cost against other models of the same type, on fal.ai list prices. Cost inside Sync.so depends on its own plans and credits.
Key Features
- sync-3 lipsync model, up to 4K at 60fps, multi-speaker and low-light handling
- Lipsync on existing footage (no avatar or shot regeneration)
- Translate and dub: translation, voice cloning and lipsync in one step, 29 languages
- Image-to-video: one still plus audio becomes a talking video
- Timeline editor with split, trim, per-segment sync and multi-track layering
- Assistant mode: describe the job, the model and speakers are picked for you
- Official MCP server, Python and TypeScript SDKs, REST API with webhooks
- DaVinci Resolve Studio plugin, Adobe Premiere plugin and ComfyUI node
- Batch API on the Scale plan
When to reach for it — and when to skip
Reach for it when…
- Fixes a performance or a line without reshooting or regenerating the shot
- Clear per-second price with no credit system to predict
- Dubbing and lipsync share one workflow, so a redub does not need a second tool
- Developer-friendly: MCP server, SDKs and webhooks are documented
Skip it when…
- Lipsync only: no generation, no shot creation, no soundtrack or music
- Free tier output is watermarked and capped at 20 seconds
- Dubbing covers 29 languages, well short of the 175+ on HeyGen
Best For
✓ Ideal for
- Redubbing a finished spot or doc for several language markets
- Fixing a flubbed line or matching new VO to an existing take
- Lip syncing an AI-generated or photo-based character to a recorded performance
- Post houses and agencies that need lipsync inside an NLE via the Resolve or Premiere plugin
- Developers building a lipsync step into a Windmill, Convex or custom pipeline through the API
✗ Not built for
- Making a talking-head presenter from a script with a stock avatar library (HeyGen or Synthesia are built for that)
- Generating the shot itself (use a video model, then lipsync it here)
Working Tips from Filmmakers Using Sync.so
- 01 Estimate cost before running: the API has an estimate-cost endpoint, and sync. bills per output frame, so a 25 fps source is cheaper than a 60 fps one.
- 02 Use lipsync-2 for drafts and sync-3 only for the final pass, since sync-3 costs about 2.7x lipsync-2 per second at the base rate.
- 03 Set sync_mode to cut_off when audio and video lengths differ, so you are billed for the shorter track.
- 04 Use the Resolve plugin to pick clips from the media pool rather than exporting and re-importing.
Sync.so pricing in 2026
| Plan | Price | What you get |
|---|---|---|
| Free | $0 monthly |
|
| Hobbyist | $5 monthly |
|
| Creator | $19 monthly |
|
| Growth | $49 monthly |
|
| Scale | $249 monthly |
|
| Enterprise | Custom annual |
|
Prices checked against Sync.so's pricing page on 9 Oct 2026. Vendors change plans often, so confirm before you buy.
Sync.so vs HeyGen vs D-ID
| Sync.so | HeyGen | D-ID | |
|---|---|---|---|
| Price | Free (3 gens/month, 20s max); Hobbyist $5/mo; Creator $19/mo; per-second usage on top (sync-3 $0.107 to $0.133/sec by plan) | Free (3 videos/mo, 1-min cap) · Creator $29/mo · Pro from $49/mo · Business $149/mo + $20/seat | Trial with watermark, then from $5.90/mo (Lite) |
| Infinite canvas | No | Partial | No |
| AI agent | No | Yes | Partial |
| Chat assistant | Partial | Partial | Partial |
| Team collaboration | Partial | Yes | Not confirmed |
| Text to image | No | Yes | Partial |
| Image editing | No | Partial | Not confirmed |
| Text to video | No | Yes | No |
| Image to video | Yes | Yes | Yes |
| Video to video | Yes | Partial | Partial |
| Native audio in video | No | Yes | Partial |
| Lip-sync | Yes | Yes | Yes |
| Voice / TTS | Yes | Yes | Yes |
| Music & SFX | No | Yes | No |
| Character consistency | Partial | Yes | Partial |
| Camera control | No | Partial | No |
| Custom training | No | Partial | Partial |
| Third-party models | Partial | Yes | No |
| Public API | Yes | Yes | Yes |
| MCP server | Yes | Yes | Not confirmed |
| Mobile app | Not confirmed | Yes | Partial |
| Desktop / local | No | Partial | Partial |
| Open source | No | Partial | No |
| Free tier | Yes | Yes | Partial |
| Commercial use | Not confirmed | Partial | Not confirmed |
Each column comes from that tool's own researched page. "Not confirmed" means the vendor doesn't say.
Open the full Sync.so vs HeyGen vs D-ID comparison →Use Cases
Tags
Creators Using Sync.so
Sources
Facts on this page were checked against Sync.so's own pricing, product and changelog pages on , then cross-checked with independent coverage where vendor pages were silent. Anything we couldn't confirm is marked "Not confirmed" rather than guessed.
- sync. pricing page sync.so
- sync. billing docs (per-second rates, free tier, max duration) sync.so
- sync. blog (2026 product launches) sync.so
- sync. changelog, May 2026 (dubbing, Assistant mode, MCP, sync-3) sync.so
- sync-3 launch post sync.so
- sync. workflows and dashboard rebuild (July 2026) sync.so
- sync. MCP server docs sync.so
- HeyGen pricing (comparison column) heygen.com
Spotted something out of date? Tell us and we'll re-check it.
Discussion
No comments yet — be the first.