AI video tools compared: generators, presenters and editors
Clips from a sentence, a presenter who never existed, a talk translated into ten languages in your own voice. What each tool is for, what it costs, what the free plans hide, and the rules on using real faces.

In short
- OpenAI has closed Sora: the app shut on 26 April 2026 and the API in September. Google, Runway, Kling and Luma now lead in generating clips.
- Google's Flow gives everyone 50 free credits a day for Veo 3.1 clips; videos carry an invisible SynthID mark, and a visible watermark unless you are on the top Ultra plan.
- Runway allows commercial use even on its free plan, but free videos are watermarked. Paid plans start at $12 a month billed yearly.
- For a presenter talking to camera, HeyGen and Synthesia make avatar videos from text. Using a real person's face needs that person's recorded consent.
- Free plans usually mean a watermark, short clips and, at some companies, your videos used for training. Read the licence before using a clip in an advert.
A year ago an AI video was a curiosity: eight wobbly seconds of a cat that grew a sixth leg. Now the clips hold together, the faces move their lips in time, and small businesses use them in real adverts.
There are three kinds of AI video tool, and they solve different problems. Generators turn a sentence or a picture into a short clip. Presenter tools make a person, real or invented, talk to camera from a script. Editors cut, caption and translate the video you already have.
How we compared them: from each company’s own pricing and help pages, checked on 6 October 2026. Prices are in US dollars. We have no affiliate deals with any of these companies.
The short answer
| You want to… | Use | Free option |
|---|---|---|
| make short cinematic clips from text or a photo | Google Flow or Runway | yes, daily or one-off credits, watermarked |
| make many clips cheaply | Kling | yes, watermarked, not for commercial use |
| a presenter reading a script, for training or sales | Synthesia or HeyGen | yes, short and watermarked |
| translate your own video into other languages | HeyGen | yes, 3 short videos a month |
| edit, caption and cut for social media | CapCut | yes |
First, Sora is gone
OpenAI’s Sora was the tool that made AI video famous. It no longer exists as a product: OpenAI shut the Sora app and website on 26 April 2026 and closed its API in September. If you used it, OpenAI offered a way to export your work before the closure. Any guide that still recommends it is out of date.
Generators
Google: Flow and the Gemini app
Google’s video model, Veo 3.1, lives in Flow, Google’s film-making tool. Everyone gets 50 free credits a day, which don’t roll over. A fast Veo clip costs 20 credits and a top-quality one 100, so the free allowance buys a couple of quick experiments a day. Paid Google AI plans add monthly credits:
| Plan | Per month | Extra Flow credits a month |
|---|---|---|
| Free | $0 | none (50 a day for everyone) |
| AI Plus | $4.99 | 200 |
| AI Pro | $19.99 | 1,000 |
| AI Ultra | $99.99 or $199.99 | 10,000 or 25,000 |
Clips are 4, 6 or 8 seconds. Subscribers can upscale to 1080p for free; 4K is for Ultra only. In the Gemini app itself, Veo has been replaced by Google’s newer Gemini Omni model, which makes 10-second videos for subscribers aged 18 and over.
Every Google video carries SynthID, an invisible mark that identifies it as AI-made. A visible watermark is added too, except for Ultra members in Flow.
Privacy note: in the Gemini app, “Keep Activity” is on by default, and while it is on your conversations may be used to improve Google’s models. Switch it off if that matters to you.
Runway
Runway is the professional’s favourite: its own Gen-4.5 model plus others, and editing tools around them.
| Plan | Billed yearly | Billed monthly | Credits a month |
|---|---|---|---|
| Free | $0 | $0 | 125, once |
| Standard | $12 | $15 | 625 (about 52 seconds of Gen-4.5) |
| Pro | $28 | $35 | 2,250 |
| Max | $76 | $95 | 9,500 |
Runway lets you use what you make commercially on every plan, the free one included, and you own the output. Free videos carry a watermark. Its terms allow it to use your inputs and outputs to train its models, and only enterprise customers get a no-training promise.
Kling
Kling, from the Chinese company Kuaishou, is popular for producing a lot of video for little money. Its 3.0 model makes clips of 3 to 15 seconds, in 1080p on paid plans. Paid plans start at about $7 to $10 a month for Standard, depending on promotions. The free tier is watermarked and not for commercial use.
Luma
Luma’s Dream Machine runs its own Ray models alongside Google’s Veo and Kling’s. Paid plans start at $30 a month ($300 a year) for Plus. According to Luma’s licensing page, videos from free plans are for personal use only, keep their watermark even if you upgrade later, and may be used for training; paid plans bring commercial rights and no training on your content.
Presenters and translation
Synthesia
Synthesia is built for companies: training videos, onboarding, internal news, with a presenter reading your script.
- Free: 10 minutes of video a month, 9 avatars, watermark.
- Starter: $18 a month billed yearly, or $29 monthly.
- Pro: $64 a month billed yearly, or $89 monthly, with up to five personal avatars.
Two rules worth knowing. Synthesia’s stock avatars may not be used in paid adverts on TV or social media. And a personal avatar, a digital copy of a real person, needs that adult’s informed consent, recorded live. Synthesia says it does not use customer data to pre-train its models.
HeyGen
HeyGen does avatars too, but it is best known for translation: film yourself once and get the same video in other languages, in your own voice, with lip movements to match.
- Free: 3 videos a month, up to a minute each, watermarked.
- Creator: $24 a month billed yearly, or $29 monthly. Videos up to 30 minutes, more than 175 languages.
- Pro: $49 a month, with 4K.
- Business: $149 a month plus $20 per extra seat.
To make a video twin of a person, that person has to record a consent video. HeyGen’s own pages are inconsistent on whether ordinary users’ data can be used for training, so check the privacy settings in your account.
Editing: CapCut
CapCut, from TikTok’s owner ByteDance, is the editor most social-media creators already use. Its AI features write captions, remove backgrounds, generate clips and offer more than 100 ready-made avatars. CapCut Pro costs $19.99 a month or $179.99 a year, and prices vary by country.
For editing interviews and talking-head videos by editing the transcript, see Descript in our audio tools comparison.
Rules for any AI video
- Never put a real person’s face or voice in a video without their consent. That includes colleagues, customers and public figures. The serious tools require recorded consent for a reason.
- Check the licence before you advertise. Free plans often forbid commercial use, and some stock avatars are banned from paid ads.
- Label it. Viewers increasingly expect to know when a person on screen isn’t real, and the EU’s AI Act requires deepfakes to be disclosed.
- Don’t remove watermarks or provenance marks. They protect you as well as the viewer.
- Budget in credits, not plans. A plan’s price says little; what matters is how many seconds of the quality you need it buys. Try the free credits first.
More tool comparisons
Sources
- OpenAI Help: What to know about the Sora discontinuation
- Google: Gemini subscriptions (checked 6 October 2026)
- Google Flow Help: Credits
- Google: Video generation in the Gemini app
- Google DeepMind: Veo
- Runway: Pricing (checked 6 October 2026)
- Runway: Terms of use
- Kling: Kling Video 3.0 credit cost guide (28 July 2026)
- Luma: Pricing
- Luma: Licensing
- HeyGen: Pricing
- HeyGen Help: Recording your consent video
- Synthesia: Pricing
- Synthesia Help: Video licensing
- Synthesia: AI governance practices
- CapCut: Standard vs Pro (23 September 2026)
More guides
- AI presentation tools compared: Gamma, Beautiful.ai, Copilot, Gemini and Canva
- AI meeting note-takers compared: Otter, Fireflies, Fathom, Granola, MeetGeek, Notuly and the built-ins
- We gave one meeting to four AI note-takers. Two knew who had to do what
- We gave four AI slide makers the same document. Two of them made up the numbers
- AI audio tools compared: voice, clean-up, editing and music
- We cleaned one noisy 35-minute recording with Diffio and Adobe. Background voices decided it
- AI data analysis tools compared: from spreadsheets to big data
- AI image tools compared: ChatGPT, Gemini, Midjourney, Firefly, Ideogram, Canva and more
Comments
What do you think? No account needed. Please be kind; comments that attack people are removed.
No comments yet. Be the first.