Sync Labs review, pricing and verdict

Last updated: 2026-09-04

Sync Labs is a developer API for lip sync and video dubbing, offering five models that range from a fast general-purpose default to sync-3, its 4K-native flagship that auto-detects hands, microphones, and glasses blocking the mouth. It targets developers wiring dubbing into apps via REST and SDKs, not end-user video editors.

About Sync Labs

Sync Labs (sync.so) is a lip sync and visual dubbing API built by Synchronicity Labs Inc. for developers, giving programmatic access to lip sync models through a REST API and Python/TypeScript SDKs, plus a web-based Studio playground for testing before writing code. The company positions it as "studio grade visual dubbing with perfect lipsync".

The platform offers five lip sync models spanning a quality and speed trade-off. sync-3 is the flagship, badged 4K and "intelligent": native 4K output with built-in obstruction detection for hands, microphones and glasses, handling extreme camera angles and full-shot processing without manual masking. lipsync-2-pro adds diffusion-based super-resolution for fine facial detail like beards and teeth. lipsync-2 is the fast, general-purpose HD default, and lipsync-1.9.0 is the legacy budget option. react-1, badged expressive and experimental, layers facial expression and head motion on top of standard lip movement, capped to short clips per generation.

Billing is the part worth reading twice. Sync Labs describes it as "subscription + usage": the subscription unlocks duration, concurrency, voice-clone slots and premium features, and usage is charged separately. The unit is not the second — it is the output frame. Sync Labs states that it counts the frames in the output video and multiplies by the model's per-frame price, that the published per-second rates are quoted at a 25 fps reference, and that "actual billing is per output frame regardless of the input frame rate". A 30 fps render therefore costs about 20% more per second of finished video than the quoted rate. Charges are invoiced once they cross a plan-specific threshold ($6 Hobbyist, $20 Creator, $50 Growth, $250 Scale), so no prepaid credits are involved.

The free tier is a trial, not a permanent allowance. New accounts get 3 free generations per month plus 10 text-to-speech generations with no credit card, capped at 20 seconds each, and at most one of the three may use sync-3 at a 15-second limit. Sync Labs states plainly that "free generations are only available on free accounts. Once you subscribe, you move to usage-based billing; there's no monthly free generation allowance on paid plans." The pricing page's plan selector starts at Hobbyist and does not show the free tier at all, which only the FAQ and the billing docs describe.

Developer tooling includes a REST API at api.sync.so, a Python SDK (pip install syncsdk), a TypeScript SDK (npm i @sync.so/sdk), a documented OpenAPI 3.1 spec and a ComfyUI plugin. Subscription-gated features named in the docs include reasoning mode, automatic speaker detection, bring-your-own TTS key, the Batch API and team workspaces.

Screenshots

Pricing

Subscription plus usage, billed per output frame. Free tier: 3 generations/month (which may include lipsync-2-pro) plus 10 text-to-speech generations, 20-second maximum, at most one of the three on sync-3 at a 15-second limit, watermarked, no credit card — and it disappears the moment you subscribe, since Sync Labs states there is no monthly free generation allowance on paid plans. Hobbyist $5/mo (1 min max, 1 concurrent, 3 voice clones, watermark).

Creator $19/mo (5 min, 3 concurrent, 5 voice clones, no watermark, own TTS API key, automatic speaker detection, react-1). Growth $49/mo (10 min, 6 concurrent, 15 voice clones, 5% usage discount, Batch API, 3 seats). Scale $249/mo (30 min, 15 concurrent, 50 voice clones, 20% usage discount, Batch API, 5 seats, delegated support).

Enterprise custom. 167/sec. Because billing counts output frames rather than seconds, rendering above 25 fps costs proportionally more than the quoted per-second figure.

Usage is invoiced at plan thresholds of $6, $20, $50 and $250 — no credits, no prepayment.

Key Features

• Five-model lip sync family: sync-3 (4K, obstruction-aware, badged intelligent), lipsync-2-pro (4K, diffusion super-resolution), lipsync-2 (HD general-purpose default), lipsync-1.9.0 (legacy budget) and react-1 (expressive, badged experimental) — each at its own per-frame rate so the quality/cost trade-off is a per-job choice.
• 4K native output with obstruction detection: sync-3 outputs native 4K and automatically detects hands, microphones and glasses blocking the mouth, handling extreme angles and full shots without manual masking. On lipsync-2 and lipsync-2-pro obstruction detection is optional rather than built in.
• Per-frame usage billing with invoice thresholds: Sync Labs counts frames in the output video and multiplies by the model's per-frame price; the published per-second rates are quoted at 25 fps, and billing is "per output frame regardless of the input frame rate". Charges invoice at $6/$20/$50/$250 by plan — no credits, no prepayment.
• Developer-first: API, SDKs, OpenAPI spec, ComfyUI: REST API at api.sync.so, typed Python (syncsdk) and TypeScript (@sync.so/sdk) SDKs, a published OpenAPI 3.1 spec, a ComfyUI plugin and a Studio playground for testing before writing code.
• Visual dubbing, voice cloning and speaker detection: react-1 layers facial expression and head motion; voice cloning scales from 3 slots on Hobbyist to 50 on Scale; automatic speaker detection, reasoning mode and bring-your-own TTS key unlock from Creator upward, with team workspaces on Growth and Scale.
• Unlimited generation, gated by concurrency: Sync Labs states all plans support unlimited video generation — concurrency limits how many render simultaneously, from 1 job on Hobbyist to 15 on Scale, and anything else queues.

Pros

• Five models at five different rates put the quality/cost trade-off in the caller's hands per job — $0.02/sec on the legacy model up to $0.167/sec on react-1, rather than one price for everything
• No credits and no prepayment: usage accrues and invoices at a plan threshold ($6 to $250), so an idle month costs only the subscription and a busy one does not require topping up a balance mid-render
• Developer-first in a category that mostly is not: typed Python and TypeScript SDKs, a published OpenAPI 3.1 spec, a ComfyUI plugin and a Studio playground for testing before writing code
• sync-3's obstruction detection handles hands, microphones and glasses over the mouth without manual masking — something competing lip-sync APIs generally leave to post-processing
• Unlimited generation on every plan, with concurrency rather than a monthly cap as the limiter, so a burst of work queues instead of being refused

Cons

• Billing is per output frame, not per second, and the published rates are quoted at a 25 fps reference. Render at 30 fps and the same finished minute costs about 20% more than the headline figure implies — the pricing page's own estimator makes fps an input for exactly this reason
• The free tier is a trial that vanishes on upgrade: Sync Labs states "free generations are only available on free accounts... there's no monthly free generation allowance on paid plans", and the pricing page's plan selector does not show the free tier at all, so only the FAQ and billing docs describe it
• Sync Labs' privacy policy says personal information is used "for developing, training, and fine-tuning the AI models we use to support the Services", and it does not separately state whether uploaded video and audio are excluded from that — an absence, not an assurance, on a product whose input is people's faces and voices
• No published compliance posture: no SOC 2, ISO or trust centre is claimed anywhere on the site, and the privacy policy offers only that Sync Labs uses "security measures that are designed to protect against unauthorized access"
• Two charges rather than one: the subscription sets the limits and the usage sets the bill, so total cost is less predictable than a flat price, and the cheapest paid tier still watermarks output — removing it requires stepping up to Creator

Visit Sync Labs Official Website