If you are adding avatar video to a product, the API price sets the floor for your gross margin. Comparing providers is harder than it should be: one sells a prepaid wallet, another a monthly plan with included minutes, a third tokens that never expire. This post puts the public numbers side by side and works through what a video actually costs at three different volumes.
All competitor prices below are public list prices as of October 2026; check the provider's site, because plans and rates change. Where a provider is better for reasons other than price, we say so.
The three pricing models
Before comparing numbers, understand what kind of bill you are signing up for.
Prepaid wallet, pay per minute
You load money in advance and each render draws it down. There is no monthly minimum, and cost scales linearly with usage. HeyGen's API works this way.
Monthly plan with included minutes
You pay a fixed fee each month for a bundle of minutes, then overage per minute. This is simple to budget at steady volume, but you pay for unused minutes in quiet months and need to pick the right tier. Synthesia and Tavus use variations of this model.
Prepaid tokens that never expire
You buy packs of tokens and each second of video uses a fixed number of them. Unused tokens roll over indefinitely, so there is no wasted spend in a slow month. PitchTwin uses this model.
The public numbers
HeyGen API
Per G2's HeyGen API pricing article:
- Avatar IV digital twin (avatar from video): $4.00 per minute, about $0.0667 per second.
- Photo avatar: $3.00 per minute, about $0.05 per second.
- Billed from a prepaid API wallet.
HeyGen also offers a full studio editor and stock avatars, which matters if your users build videos by hand.
Synthesia
Per this 2026 Synthesia pricing breakdown:
- API access starts on the Creator plan at $89 per month with 30 video minutes, about $2.97 per minute.
- Enterprise includes 360 API minutes per year.
- Not every avatar type is API-compatible.
Synthesia's strength is its polished platform and library of hundreds of stock avatars, well suited to corporate training teams.
Tavus
Per the Tavus pricing page:
- Monthly developer plans, with tiers listed at $59, $397 and $975 per month among others.
- Plans bundle real-time conversational minutes and async video generation minutes.
- Video generation overage runs from about $1.00 per minute on lower tiers to about $0.80 per minute on higher ones.
Tavus is the one to look at first if you need real-time conversational video, where an avatar talks with a user live.
PitchTwin
From the PitchTwin pricing page:
- 1 token per second of video when you bring your own audio; 1.5 tokens per second when PitchTwin voices the script in the avatar's cloned voice.
- Packs: Starter $9 for 600 tokens, Growth $39 for 3,000, Scale $149 for 12,000.
- Effective per minute: $0.90 / $0.78 / $0.74 with your own audio, $1.35 / $1.17 / $1.12 with PitchTwin's voice.
- 60 free tokens on signup with no card, no seats, no subscription. Tokens never expire, and failed generations are refunded automatically.
PitchTwin does not offer a studio editor, stock avatars or real-time video. Every avatar is built from a real video of the person.
Summary table
| Provider | Model | Entry point | List price per minute |
|---|---|---|---|
| HeyGen API | Prepaid wallet | Wallet top-up | $4.00 (digital twin), $3.00 (photo) |
| Synthesia | Monthly plan | $89/mo (Creator, 30 min) | ≈ $2.97 within Creator |
| Tavus | Monthly plan + overage | $59/mo tier | ≈ $1.00 → $0.80 overage |
| PitchTwin | Prepaid tokens | $9 pack | $0.74–$0.90 (your audio), $1.12–$1.35 (voiced) |
Public list prices as of October 2026; check the provider's site.
Worked examples
Per-minute rates only matter when you multiply them by your real usage. Here are three scenarios. Figures are list prices before any negotiated discount.
Scenario 1: an early-stage product, 20 minutes a month
Twenty one-minute demo videos, voiced by the provider.
- PitchTwin: 20 minutes × 90 tokens = 1,800 tokens. One Growth pack ($39, 3,000 tokens) covers it, and the remaining 1,200 tokens carry over to next month.
- HeyGen digital twin: 20 × $4.00 = $80.
- Synthesia Creator: $89 for the month, with 10 of the 30 included minutes unused.
- Tavus: the plan fee for whichever tier fits, with included video minutes depending on tier.
At this stage the absolute numbers are small. What matters more is that you are not locked into a monthly fee before you have customers.
Scenario 2: a growing B2B tool, 375 minutes a month
500 personalized videos of 45 seconds each.
- PitchTwin, voiced: 375 × 90 = 33,750 tokens, about $419 at Scale-pack rates.
- PitchTwin, your own audio: 375 × 60 = 22,500 tokens, about $279 at Scale-pack rates, plus whatever you pay for speech elsewhere.
- HeyGen digital twin: 375 × $4.00 = $1,500.
- Synthesia: exceeds the Creator plan's 30 minutes; you would need a larger plan.
- Tavus: plan fee plus overage beyond the included minutes; 375 minutes of overage alone at $0.80–$1.00 is $300–$375.
Scenario 3: high volume, 50,000 minutes a month
100,000 videos of 30 seconds each.
- PitchTwin: 3,000,000 tokens with your own audio ($37,250 in Scale packs) or 4,500,000 tokens voiced ($55,875).
- HeyGen digital twin: 50,000 × $4.00 = $200,000.
- Synthesia and Tavus: at this volume you are in enterprise or custom pricing territory.
At this scale every provider, PitchTwin included, will discuss volume pricing, so use list prices to decide who to talk to, not as the final number. For a design that handles this volume, see architecture for 100,000 personalized videos a month.
Hidden costs to check
The per-minute rate is only part of the bill. Ask each provider:
- Is the voice included? A per-minute price that includes text-to-speech is not comparable to one that requires you to bring audio. PitchTwin shows both rates separately; if you already pay for speech, see using your own ElevenLabs audio.
- Do unused minutes expire? Monthly plans reset. Prepaid tokens that never expire do not.
- Are failed renders charged? At scale, even a small failure rate adds up. PitchTwin refunds failed generations in full automatically.
- Are there seat fees? Some platforms price per editor seat. An API used by your backend should not need seats.
- Is the avatar type you need available through the API? Not every avatar type on every platform is.
- Is there a watermark or provider branding? If you resell video under your own brand, branding on the output is a hidden cost you pay in trust.
- What happens on an engine upgrade? A model update that changes how a face looks can force you to re-render or re-approve content. PitchTwin pins each avatar to its engine version.
How to compare for your own product
Build a small spreadsheet with four inputs: average video length, videos per month, whether you need the provider's voice, and expected month-to-month variance. Then compute monthly cost for each provider using its actual pricing model, not just the headline per-minute rate. Divide by the number of videos to get cost per video, and compare that to what you charge your customer for it.
Price is one input. If your users need a studio, stock avatars or live conversation, a cheaper API that lacks them is not cheaper for you.
Try it
You can check quality before you think about volume: sign up for 60 free seconds with no card and render a test from your own footage. The docs show the full request format, and the pricing page lists current packs.