B2B API for AI avatar video
Building a B2B tool that generates avatar videos? PitchTwin is the API behind it. Avatars made from a real video of the person look 1:1 like them, not plasticine. White-label, from $0.74 a minute.
Each avatar is one real person, from a clip you upload or a 5-minute call with our AI host: same face, any script, any language.
Each rep records once. Your tool sends a personal video in every sequence step.
text → videoRead the guideCreators license their face once. Your users spin out hundreds of ad variants.
script + pausesRead the guideTurn every course update into a new video from the same instructor, in any language.
multi-languageRead the guideThe customer's own CSM greets every new account by name, automatically.
webhooksRead the guideRecruiters answer every candidate on camera at every pipeline stage.
text → videoRead the guideBring your own translated audio. The original speaker says it, on camera, in the new language.
your audioRead the guideBuilding something else with talking avatars? Tell us what.
Most avatar APIs re-generate the whole face, and your users notice the waxy skin and frozen eyes. We start from real video of the person and keep it.
Use the footage you already have, or let Lily, our AI host, get it for you. Both give the same 1:1 avatar and the same avatar_id.
Already have video of the person? Send 30 seconds to 2 minutes of them talking to camera.
Your user opens a link and talks to Lily, an AI host under your brand and your name for her. Nothing to film, no script to read, no file to upload.
Tell her what your product does and what to find out. For a sales tool she asks what they sell and to whom. For an ad generator, who the audience is and how they talk. You get it back as a ready text profile and a full transcript, to reuse in every script you write for that person.
Put in your users' volume and what your current provider charges. The difference is yours.
3,750 minutes of video. Scale-pack rates: $0.74/min with your audio, $1.12/min when we voice it.
curl https://pitchtwin.ai/api/v1/generations \ -H "Authorization: Bearer $PITCHTWIN_KEY" \ -d '{ "avatar_id": "av_maya", "script": [ { "text": "Hi Sarah, saw Acme just raised —" }, { "text": "congrats!", "pause_after_ms": 300 } ], "webhook_url": "https://your.app/hooks/video" }' # 202 Accepted { "id": "gen_41c…", "status": "queued" }
Prepaid tokens that never expire. One token is a second of video with your audio, 1.5 when we voice it. 60 free to start. Full pricing
For products past 10,000 minutes a month: committed pricing, invoicing, custom limits, a shared Slack.
Talk to the foundersFor most avatar products the video API is the biggest line in the cost of every sale. Here's what a minute costs.
| API | Avatar from a real person's video | Price per minute | Monthly plan needed |
|---|---|---|---|
| PitchTwin | Yes, kept 1:1 from the clip | $0.74–1.35 | No. Prepaid, tokens never expire |
| HeyGen API | Yes, digital twin | $4.00 | No, prepaid API wallet |
| Synthesia | Yes, personal avatar | ≈ $3.00 | Creator plan, $89/mo, for API access |
| Tavus | Yes, replica | $0.80–1.00 + plan | Monthly plan from $59/mo |
Public list prices as of October 2026, for avatar video made over the API. HeyGen: Avatar IV digital twin $4.00/min, photo avatar $3.00/min. Synthesia: API from the Creator plan. Tavus: per-minute overage on top of a monthly plan. Check each provider's site for current terms.
Anything else? Email us.
Yes, that's what it's for. Videos carry no PitchTwin branding and land on your webhook. Your customers never need to know we exist.
No. You create avatars and videos with your own API key, one avatar per person in your product. You decide how to bill your users.
We build every avatar from real video of the person and keep it, so the result looks like them 1:1 instead of a smooth generated face. It also costs several times less per minute, with no subscription and no seats, and each avatar keeps its engine version so faces never shift under your users.
Thirty seconds to two minutes of one person facing the camera in steady light. A phone is enough. With 10+ seconds of speech we clone the voice from the clip too.
Our voice engine speaks dozens of languages, set per request. Or send your own audio in any language and the avatar speaks it.
Everyone in an avatar must have agreed to it. Your terms should say so; ours require it. Avatars and their cloned voices can be deleted at any time through the API.