The short answer
Synthesia is a London-based AI video platform that turns scripts, documents and slides into videos presented by AI avatars in 140+ languages and voices. It is built for training and internal communication. Its Express-3 avatar model and AI Assistant launched on all plans on 15 July 2026, and Sessions added two-way avatar conversations on 1 October 2026. The free Basic plan includes 10 minutes of video a month with a watermark.
| Made by | Synthesia, United Kingdom |
|---|---|
| Free plan | Yes, Basic: 10 minutes of video a month, 9 avatars, watermark |
| Models | Express-3 avatars, Dubbing 2.0 |
| Best at | Training and onboarding videos, easy updates, localisation, roleplay practice |
| Shows sources | Not applicable: it generates video |
| Team plans | Enterprise, with SSO, workspaces, SCORM and live collaboration |
| Your data | Synthesia says customer inputs and outputs are not used to pre-train its AI |
What Synthesia is in late 2026
Synthesia, based in London, is an AI video platform for businesses. You turn a script, a document, slides or a web page into a video presented by an AI avatar, in 140+ languages and voices, and you update it later by editing the text instead of reshooting. Its home ground is training, onboarding, compliance and internal communication, and in January 2026 it raised a $200 million Series E at a $4 billion valuation to focus on how companies train their people.
Three releases shaped the product this year. On 15 July 2026 Synthesia launched an AI Assistant and the Express-3 avatar model. Express-3 improves lip sync and ties gestures to the script, lets avatars show emotion that matches what they are saying, such as serious or friendly, and generates videos up to twice as fast. It is rolling out in stages, starting with new stock avatars. On 1 October 2026 came Sessions, where avatars hold two-way conversations. And on 9 October 2026 Synthesia launched Syren in beta, an agent that builds videos with motion graphics from a prompt or a document.
Plans changed on 1 October 2026 too. The Creator plan was renamed Pro at the same price, and every plan, including the free Basic plan, now includes 10 minutes of video or dubbing, 3 survey responses and 500 credits each month. Larger Pro 2x, 4x, 6x and 8x tiers scale up to 48,500 credits a month.
What Synthesia is best at
Training videos that are easy to keep current
The real cost of training video is not making it once but updating it every time a policy or a screen changes. In Synthesia you change a sentence in the script and render again, with the same avatar and branding. The AI Assistant, available on all plans, turns a prompt, files or URLs into a structured, scripted video. You can plan the structure, tone and number of scenes in chat first, pick a Cinematic or Presentation style, and it applies your brand kit automatically if you have one (the pricing page lists brand kits on Enterprise).
Videos from documents and slides
Syren takes a text prompt plus branded documents, images, videos, PowerPoint files or YouTube links and builds motion graphics, animation, avatars, voiceover, music and b-roll, then refines the result with you in chat. Videos run up to three minutes, horizontal or vertical. You find it under the Syren button in the left-hand menu. It is credit-based and available on free, self-serve and Enterprise plans, and Enterprise admins can switch it off for the organisation.
Practice conversations and interviews
Roleplay Sessions let people practise sales discovery, objection handling, difficult manager conversations or support escalations with an avatar that pushes back, with real-time coaching and scoring. Survey Sessions have an avatar interview people one to one and return structured insights, which works for new-hire check-ins and customer research. Expert Sessions, an avatar trained on your product that guides users step by step, is announced but not yet released.
Localisation
Self-serve plans can dub into 70+ languages and Enterprise into 140+. Dubbing 2.0, released on 15 July 2026, improved lip sync through fast cuts and multi-speaker scenes and made voices sound more natural. Enterprise also gets 1-click translation of a whole video.
Where Synthesia falls short
- The best team features are Enterprise only. SCORM export for your learning system, 1-click translation, voice cloning, dialogue scenes, live collaboration, workspaces and SSO all sit on Enterprise.
- The free plan is limited. Basic has 9 stock avatars, no personal avatar, one editor and a Synthesia watermark.
- Desktop for editing. Self-serve users can do some things in a mobile browser, but the full video editor needs a desktop or laptop, and mobile is not supported for Enterprise users.
- Not built for creative social video. For cinematic shots or playful Reels, tools such as Luma or Pika are a better fit.
- Syren is in beta. It is capped at three-minute videos for now.
How to get better results from Synthesia
- Start from your source material. Give the AI Assistant or Syren the actual policy, slide deck or process document rather than describing it. The script will be more accurate.
- Agree the structure in chat first. Set the audience, length, tone and number of scenes before the Assistant builds anything.
- Keep scenes short. One idea per scene makes later edits easy, because you change one scene instead of re-timing a whole video.
- Write for listening. Short sentences and spoken words. Read it aloud before you render.
- On Enterprise, set up your brand kit once. The Assistant then applies colours, fonts and logos for you on every video. On self-serve plans, paste your colours and tone into the prompt instead.
- Check every translation with a native speaker before it goes to a whole region, especially for compliance content.
Privacy, consent and governance
Synthesia says it does not use customer data, including inputs or outputs, to pre-train its AI components. Any fine-tuning with customer data follows the customer's written instructions and governing agreement. It lists ISO 42001 certification for responsible AI and SOC 2 compliance, consent safeguards for custom avatars and voices, and in July 2026 it was among the first AI video companies to sign the EU's Article 50 Code of Practice on AI-generated content. That makes it one of the easier tools to get past an IT or legal review. You still need written consent from any employee whose likeness becomes an avatar, and a rule for who may use it.
Which Synthesia plan suits whom
Basic is free and enough to try the AI Assistant and Syren, but every video carries a watermark. Starter removes the watermark, adds MP4 downloads, 125+ stock avatars and 3 personal avatars, and suits one person making occasional videos. Pro (formerly Creator) adds 180+ avatars, 5 personal avatars, branded video pages and interactive videos, and the Pro 2x to 8x tiers add credits for heavier use. Learning and development teams that need SCORM export, SSO, collaboration and wide translation should go straight to Enterprise. Check the official pricing page for current prices.
Synthesia is the right pick for HR, learning, compliance and internal communications teams who make many structured videos and must keep them up to date. If you want lifelike presenters for marketing or fast dubbing of existing videos, compare it with HeyGen in HeyGen vs Synthesia, see the Synthesia alternatives and start with our Synthesia lessons.
Prompts to copy
Policy document to training video
Turn this policy into a three-minute training video for [audience]: [paste or upload policy]. Five scenes: why it matters, the three rules, one realistic example, what to do if unsure, and a recap. Friendly but serious tone, Presentation style.
You get: A structured training script the Assistant can build straight away
Slides to explainer with Syren
Using this deck [upload PowerPoint], make a vertical two-minute explainer for new joiners about [topic]. Use our brand colours, animate the key numbers on slides 3 and 5, and add calm background music and an avatar presenter for the opening and closing.
You get: A short branded explainer built from slides you already have
Design a sales roleplay
Design a Roleplay Session for a [role] practising [scenario, for example a pricing objection from a procurement lead]. Describe the avatar's persona, three objections it should raise in order, what a strong answer includes, and how to score the learner out of 10.
You get: A realistic practice scenario with clear scoring
New-joiner check-in survey
Write the questions for a Survey Session that checks in with new hires after 30 days. Six open questions about onboarding, tools, their manager and workload, each with one follow-up question the avatar should ask if the answer is short.
You get: An interview script that returns usable feedback
Update an existing video
Our process changed: [describe change]. Here is the current script: [paste]. Rewrite only the scenes affected, keep everything else word for word, and list which scene numbers changed so I can re-render just those.
You get: A minimal update instead of a new video
Prepare a script for translation
Review this script before we dub it into [languages]: [paste]. Flag idioms, jokes, local references and units that will not translate well, suggest neutral alternatives, and list product names that must stay in English.
You get: A script that translates cleanly into every language