The short answer
HeyGen is an AI video platform from Los Angeles-based HeyGen that turns a script into a video of a realistic presenter, either a stock avatar or your digital twin. Its Avatar V model, announced in April 2026, builds a twin from about 15 seconds of phone video. It also translates videos into 175+ languages with lip sync on paid plans, while the free plan covers 30+. The free plan allows 3 watermarked videos a month, each up to 1 minute.
| Made by | HeyGen, United States |
|---|---|
| Free plan | Yes, 3 watermarked videos a month up to 1 minute |
| Models | Avatar V, Avatar IV, Avatar III, HeyGen Video, plus Veo 3.1 and Seedance 2.5 on Pro |
| Best at | Presenter videos from a script, video translation with lip sync, quick explainers |
| Shows sources | Not applicable: it generates video |
| Team plans | Business and Enterprise, with SSO and shared workspaces |
| Your data | May be used for training on non-enterprise plans unless you opt out by email |
What HeyGen is in late 2026
HeyGen, based in Los Angeles, makes AI avatar videos. You write or paste a script, choose a presenter, and a realistic person delivers it on screen with a natural voice. The presenter can be one of more than 700 stock avatars on paid plans or a digital twin of you. HeyGen has grown well beyond that core, though: it now translates existing videos with lip sync, builds complete videos from a single prompt through Video Agent, and has its own general video model.
The model that matters most is Avatar V, announced in April 2026 and described by HeyGen as its most advanced avatar model. It builds your digital twin from a single phone recording of about 15 seconds, with no studio or lighting setup, and keeps your face consistent across angles, camera moves and long videos. Record once and you can change the outfit, setting or look while keeping your own movements and expressions. Since July 2026 HeyGen can generate a single 30-minute avatar video in one pass on Avatar III, IV and V.
On 21 October 2026 HeyGen added HeyGen Video, a general model that generates the picture, dialogue and sound effects of a scene in one pass from text, an image or a reference, in styles such as live action, claymation and anime. You can try it free with a HeyGen account: the playground allows 2 clips a day of 5 seconds at 768p, and the app offers longer clips and 2K.
What HeyGen is best at
Presenter videos without a camera
Product updates, onboarding messages, sales follow-ups and course lessons all work well as a person talking to camera, and HeyGen lets you make them from a script. Once your digital twin exists, a new video is a matter of editing text, so a weekly update takes minutes rather than a recording session.
Translating videos you already have
On paid plans, Video Translation dubs a video into 175+ languages and dialects (the free plan covers 30+). You import a file (MP4, MOV or WEBM up to 5 GB) or paste a YouTube or Google Drive link, choose source and target languages (up to 10 per job), then pick an engine: Audio Only re-voices without lip sync, Speed adds lip sync for front-facing speakers, and Precision handles speaker switches and side profiles best. Under Advanced you can add a brand glossary so product names stay correct. Proofread, which lets you edit the translated script before the final render, is on Pro, Business, Team and Enterprise.
Whole videos from one prompt
Since July 2026, Video Agent includes HyperFrames, so one prompt can produce the script, voiceover, avatar presenter, captions, music and motion graphics. Its Storyboard mode proposes five angles and sketches key frames for your approval before rendering. Related tools turn a pasted website URL into a launch video using the site's own screens, fonts and colours, and Video Podcast turns a script, URL, PDF or document into a two-host video podcast.
Making videos on your phone
HeyGen has apps for iOS and, since its May 2026 release, the full app on Android, so you can record your avatar footage, generate and share without a laptop.
Where HeyGen falls short
- The free plan is very small. You get 3 videos a month, each up to 1 minute, with a watermark.
- On-screen text is not translated. Video Translation changes the speech, not graphics or burned-in captions. HeyGen's help centre suggests rebuilding those videos in HeyGen Studio instead.
- Lip sync has limits. It works best with one speaker at a time, facing the camera within about 45 degrees. Music and singing give inconsistent results.
- Credits vary. Credits are shared across features and the cost depends on model, length and complexity. Precision translation, for example, costs 10 credits a minute against 4 for Audio Only.
- Avatars still feel like avatars. For a sensitive message, such as layoffs or an apology, a real person on camera is the better choice.
How to get better results from HeyGen
- Record your twin carefully. Use good daylight, a quiet room and a steady phone at eye level. The 15-second clip is all the model learns from, so a clean recording pays off in every video afterwards.
- Write for the ear. Short sentences, one idea each, and no lists read aloud. Read the script out loud once before you generate.
- Test one language first. HeyGen's own advice is to translate into one language, check it, then run the rest.
- Use the brand glossary. Add product and company names so translations do not mangle them.
- Approve the storyboard. In Video Agent, check the proposed angles and key frames before rendering so you do not spend credits on the wrong idea.
- Disclose it. Tell viewers when a presenter is an AI avatar. It protects trust, and in many workplaces it is policy.
Privacy settings and your likeness
HeyGen's privacy policy, updated on 15 October 2026, says it may use data to train and improve its models, including its avatar creation models. There is no in-app switch: you opt out by contacting the privacy address listed in the policy, which is worth doing before you upload your face and voice. The policy does not cover enterprise and business offerings, which are governed by customer contracts, and HeyGen's trust page says enterprise customer data is not used to train its AI. Consent is built in for digital twins: HeyGen will not create a digital twin until the person shown records their own on-camera consent video, submitted with the footage through a QR code on their phone. Photo avatars have no such check, and HeyGen's trust page says you must have the legal rights and explicit consent of anyone whose likeness you use, so getting and keeping that agreement is your job. Decide in advance who in your organisation may create videos with a person's twin.
Which HeyGen plan suits whom
The free plan is enough to test the idea with your own avatar. Creator removes the watermark, allows videos up to 30 minutes at 1080p and includes limited Avatar V, which suits a solo creator or a manager making regular updates. Pro adds 4K, full Avatar V, newer models such as Veo 3.1 and Seedance 2.5, and the highest usage limits, and it includes translation proofreading. Business is the first plan for teams, with a shared workspace, SAML/SSO, a brand kit, SCORM and LMS export for training teams, 60-minute videos and connections to Zapier, HubSpot, Make and n8n. Enterprise adds SCIM, MFA, custom terms and no length limit. Check the official pricing page for current prices.
HeyGen is the stronger pick for marketing, sales and creators who want lifelike presenters and fast translation. Learning teams that care most about structured training content should also look at Synthesia; see HeyGen vs Synthesia, the best AI for video and our HeyGen lessons.
Prompts to copy
Turn notes into a presenter script
Turn these notes into a 60-second script for my HeyGen avatar to read to [audience]: [paste notes]. Short spoken sentences, one idea each, a friendly opening line, and end with one clear action. Mark where to pause with [pause].
You get: A script that sounds natural when an avatar reads it
Brief Video Agent
Make a 90-second explainer video about [topic] for [audience]. Use my avatar as presenter, add captions, calm background music and simple motion graphics for the three key numbers: [numbers]. Show me the storyboard before rendering.
You get: A full video plan you approve before any credits are spent
Website to launch video
Create a 45-second launch video from [URL]. Use the site's screens, fonts and colours, open with the main benefit in one sentence, show three features, and end with the sign-up page and the line [call to action].
You get: A launch video that matches your brand without design work
Glossary for translation
Here is our script: [paste]. List every product name, feature name, acronym and person's name that must not be translated, and give the correct pronunciation for each, so I can add them to HeyGen's brand glossary.
You get: A glossary that keeps translated videos accurate
Personal sales follow-up
Write a 30-second follow-up video script for [prospect name] at [company] after our call about [topic]. Mention the two things they cared about: [points]. Keep it warm, not salesy, and end by proposing [next step].
You get: A short personal video that stands out in an inbox
Weekly team update
Turn these bullet points into a two-minute weekly update for my team, read by my avatar: [bullets]. Open with the one thing that matters most, then wins, then blockers, then what I need from them. Plain, direct tone.
You get: A weekly video update without recording yourself