Synthesia
The avatar card. Synthesia turns a script into a video presented by a photorealistic person in any of 160-plus languages, and nine in ten Fortune 100 companies use it for training and internal video. In 2026 it added live roleplay coaching and a real-time avatar API. Here is what each plan buys in minutes, the consent rules that govern whose face you can use, and where a cheaper tool will do.

- Rating8.1/10
- Read time8 min
- Data tables2
- StatusChecked September 2026
Synthesia
The most polished AI presenter for business video, with the strictest consent rules in the suit and a minute allowance that runs out fast on self-serve plans.
- Best for
- Training, onboarding, product explainers, internal comms and multilingual versions of all of them, made without a camera
- Price
- Basic free; Starter $29/mo or $18/mo billed yearly; Creator $89/mo or $64/mo billed yearly; Enterprise on request, billed annually
- Free tier
- 1,200 credits a month, about 10 minutes of video, nine stock avatars, watermark, no commercial use
- Made by
- Synthesia, London, founded 2017
- Platforms
- Web app, Chrome screen-recorder extension, API from Creator up, SCORM and LMS export on Enterprise
- Trains on your data?
- Enterprise content is excluded from training per the privacy policy; the position for free and self-serve accounts is less explicit, so ask for written confirmation if it matters
Rating reflects Synthesia's documented plans and capabilities as of September 2026 plus our own sessions. Some vendor links on this site are affiliate links; they never affect a rating. See our disclosure and method.
What it does
You write or paste a script, pick an avatar and a voice, add slides, screen recordings, text and B-roll in a scene-based editor, and Synthesia renders a video of a person delivering it. That is the core. The avatars are built from footage of real actors who are paid through a talent program, and since January 2026 the higher-quality Express-2 generation has been available on every plan. Voices cover more than 160 languages and accents, translation covers 70-plus output locales on self-serve plans and 140-plus on Enterprise, where one-click translation of a whole video lives, and Dubbing 2.0, shipped in July 2026, lip-syncs translated audio across 130 languages.
Around that core Synthesia has spent 2026 building a training platform rather than a video tool. Roleplay Sessions, launched July 22, put an employee in a live conversation with an avatar that pushes back, then scores them against a rubric; TechCrunch noted the reasoning underneath is OpenAI's while the avatar and voice are Synthesia's own. The Interactive Avatar API, launched July 15 for Enterprise customers, lets developers embed a real-time, interruptible avatar on Synthesia's Express-3 model into their own product while bringing their own language model, speech-to-text and text-to-speech. July also brought brand-customizable avatars, AI video presets that generate B-roll without a prompt, motion-graphics editing with auto-translation, and more natural voice variants.
The AI Playground tab gives every plan access to third-party video models for B-roll, including Google's Veo 3.1 and, at the time of checking, OpenAI's Sora 2. OpenAI has said it is shutting the Sora API on September 24, 2026, so expect that option to change.
Where it wins
- Presenters that pass for corporate video. Memeburn's September 2026 test found lip sync held on multi-syllable words, blinks and head movement looked natural, and a personal avatar handled a script it had never been recorded saying. For a compliance module or a product walkthrough, nobody will complain.
- Languages. One script, 160-plus voices, a translation button and lip-synced dubbing. A five-language onboarding video that once took a week of studio time now takes an afternoon. No other card in the suit matches the language count.
- Consent rules that actually bite. A personal avatar requires footage plus a separate recorded consent statement, and the two faces must match or the request is rejected. A custom avatar of anyone else requires their free, informed consent and confirmation that they are of age. Stock avatars cannot appear in paid ads or broadcast TV, and opinionated or sensitive topics are restricted to avatars of yourself, to protect the actors.
- Governance you can show a lawyer. ISO 42001 certification, a published AI impact assessment, content moderation at the point of creation with human review, external red teaming and an annual report. An independent May 2026 review of six avatar platforms' privacy policies called Synthesia the benchmark and said it was not close.
- Ease. Nobody on your team needs to know what a keyframe is. Templates, a script assistant, PowerPoint import and bulk personalization from a spreadsheet cover most of what an L&D team asks for.
Where it loses
- Ten minutes a month. Starter is $29 for 10 minutes of finished video, Creator is $89 for 30. Minutes are counted by the second, do not roll over, and when you run out you wait or upgrade. If you plan to publish weekly, you are an Enterprise customer whether you meant to be or not.
- Per-seat everywhere. Each plan includes one editor. Guests can view and comment, but a second person who needs to edit needs a second subscription. A three-editor Creator team is $267 a month.
- The realism ceiling. Urgency, humor and emphasis still come out flat. Memeburn's testers fed an avatar a line meant to sound urgent and got no change in pacing or expression. Gestures repeat. For a keynote or an ad you would still hire a human, and the stock-avatar license would not let you run the ad anyway.
- Enterprise gates the interesting parts. SCORM export, single sign-on, the 240-plus avatar library, unlimited personal avatars, Roleplay at scale, the Interactive Avatar API and 140-language translation all sit behind a sales call. The Studio Express-1 avatar, the premium custom-avatar tier, is a $1,000 a year add-on on annual plans and takes up to ten days.
- Two annual prices on one page. The plan cards show Starter at $18 and Creator at $64 a month billed yearly; the FAQ on the same page quotes $264 and $804 a year, which is $22 and $67. We have printed both. Confirm the number at checkout before you budget for seats.
Plans and pricing, September 2026
| Plan | Price | What you get | Who it is for |
|---|---|---|---|
| Basic | $0 | 1,200 credits a month (about 10 minutes of video or 10 minutes of dubbing), nine stock avatars, 160-plus voices, AI Playground, script assistant, watermark, no commercial use | Testing the editor |
| Starter | $29/mo; $18/mo billed yearly on the plan cards ($264/yr in the FAQ) | 10 minutes of video a month (120 a year on annual), 48 minutes of dubbing, 125-plus avatars, 3 personal avatars, custom on-brand avatars, watermark removed, 1 editor and 3 guests | A solo trainer or founder making one short video a week |
| Creator | $89/mo; $64/mo billed yearly on the plan cards ($804/yr in the FAQ) | 30 minutes of video a month (360 a year on annual), 140 minutes of dubbing, 180-plus avatars, 5 personal avatars, API access, branded video pages, multiple avatars per scene, 1 editor and 5 guests, priority support | Small teams and agencies |
| Enterprise | Custom, billed annually | Unlimited minutes and dubbing, 240-plus avatars, unlimited personal avatars, one-click translation into 80-plus languages and 140-plus output locales, SCORM, SSO, live collaboration, shared workspaces, Roleplay Sessions, Interactive Avatar API, dedicated success manager | L&D, HR and enablement teams |
Credits are one pool across every metered feature. Lip-synced dubbing spends them at twice the rate, and each customized B-roll action costs 96 credits. A Personal plan from earlier years is in maintenance mode: you can stay on it while your subscription is active, but you cannot go back once it lapses.
Consent, checked September 2026: to make an avatar of yourself you record one to five minutes of footage in a single take, then record a scripted consent statement on camera and tick a biometric consent box. Processing takes about one business day and includes a human check that the two recordings show the same person. Deleting the avatar returns the slot to your plan.
Score breakdown
| Criterion | Score | Why |
|---|---|---|
| Capability | 8.5 | Best business presenters, 160-plus languages, lip-synced dubbing, live roleplay; emotional range is still limited |
| Ease of use | 9.0 | Scene editor, templates and translation that a non-editor can run on day one |
| Value | 6.5 | Ten minutes for $29 and one editor per plan; the useful features are Enterprise |
| Trust | 8.5 | Consent verification, ISO 42001, moderation at creation, clear licensing; training terms for self-serve could be plainer |
| Ecosystem | 8.0 | API on Creator, SCORM and LMS on Enterprise, third-party video models in the Playground, bring-your-own-stack interactive avatars; little for individual creators |
Verdict: the right card for a company, the wrong one for a creator.
If your videos are training, onboarding or explainers and they need to exist in several languages, Synthesia is the safest and easiest tool in the suit, and the consent rules are a feature rather than a nuisance. Start on Basic to test the editor, buy Starter annually if you make a short video a week, and talk to sales the moment you need SCORM, a second editor or more than 30 minutes a month. If you are a YouTuber or a marketer who wants cinematic or emotive video, look elsewhere.
Alternatives
- Descript if you are on camera yourself and want to edit the recording by editing the transcript, with your own cloned voice for fixes.
- ElevenLabs if what you need is the voice and the dubbing without the avatar.
- Canva if your explainer is one part of a wider social or slide workflow and a stock presenter is optional.
Quick answers
Is Synthesia free?
There is a free Basic plan with 1,200 credits a month, which is about 10 minutes of video, nine stock avatars and no card required. Videos carry a Synthesia watermark and cannot be used commercially. Starter at $29 a month removes the watermark and is the first plan you can publish from.
Can I make a Synthesia avatar of someone else?
Only with their documented consent. A personal avatar is created from one to five minutes of footage plus a separate on-camera consent statement, and the person in both must be the same or the submission is rejected. For a custom avatar of another person you must confirm they are over the legal age and gave free, informed consent, and you carry the licensing terms you agreed with them.
How many minutes of video do I get?
Ten minutes a month on Basic and Starter, 30 minutes on Creator, and unlimited on Enterprise. Minutes are counted by the second, do not roll over, and once you hit the cap you wait for renewal or upgrade. Dubbing has its own allowance: 48 minutes a month on Starter and 140 on Creator. Lip-synced dubbing uses credits at twice the rate.
Sources
- Synthesia — Pricing: plans, minutes, avatars, credits and the billing FAQ, checked September 2026
- Synthesia Knowledge Base — How do I create my personal avatar, consent recording and plan allowances, checked September 2026
- Synthesia Knowledge Base — Video licensing for stock and custom avatars, July 2026
- Synthesia — Interactive Avatar API launch, July 2026
- TechCrunch — Synthesia launches Roleplay Sessions, July 2026
- Memeburn — Synthesia AI review, September 2026


