Models
Every Atmee avatar runs on one of two models. Atmee V2 is our latest and most human-like; Atmee V1 is realistic at half the credits.
Every Atmee avatar is built on one of two models. The model decides how the face looks and moves, and what a minute of conversation costs. Everything else works the same on both: the persona, the knowledge, the API calls and the ways to build.
Which one? Choose Atmee V2 when the conversation has to feel like talking to a person: sales, coaching, interviews, anywhere the face carries part of the message. Choose Atmee V1 for conversations at high volume, or while your account waits for V2 access.
Atmee V2Latest
Our latest, most human-like model. Expressions, gestures and body language follow the conversation.
- Behavior-driven Full-stack
- 2 credits per minute
- Avatar mode
- 1 credit per minute
- Availability
- Growth and up, by invite
Atmee V1
Our earlier model: sharp, realistic and half the credits. Built for everyday conversations at volume.
- Behavior-driven Full-stack
- 1 credit per minute
- Avatar mode
- 0.5 credits per minute
- Availability
- Starter and up
Atmee V2
Our latest model, and the most human-like we have built.
- Expressive. Its face shows emotion that fits what it is saying, not a fixed smile.
- Natural motion. Gestures, head and body movement follow the conversation, so it never looks frozen between sentences.
- Listens like a person. It reacts while your user is talking, instead of waiting still for its turn.
- Pick the best take. When you create a V2 avatar, Atmee can capture several takes of it, and you keep the one you like.
V2 comes with the Growth plan and up, and is invite only for now: an account needs an invite code to use it, whatever its plan. Join early access at atmee.ai/v2signup; once you have a code, redeem it on the API page in Studio, or give it to your AI agent and it redeems it for you.
Atmee V1
Our earlier model, still realistic, at half the credits of V2.
- Sharp and realistic. A talking head rendered from a single portrait, with lips in sync with its voice.
- Ready fast. An avatar is ready to talk soon after you create it, with no takes to choose from.
- Built for volume. Half the price of V2 for every minute, on every plan with API access, from Starter up.
Pricing
Both models are paid in credits from the same balance, by the minute.
| Option | Model | Start with | Credits | Plans |
|---|---|---|---|---|
| Behavior-driven Full-stack | Atmee V2 | POST /v1/sessions | 2 credits per minute | Growth and up, by invite |
| Behavior-driven Full-stack | Atmee V1 | POST /v1/sessions | 1 credit per minute | Starter and up |
| Avatar mode | Atmee V2 | LiveKit Agents plugin | 1 credit per minute | Growth and up, by invite |
| Avatar mode | Atmee V1 | LiveKit Agents plugin | 0.5 credits per minute | Starter and up |
One credit is one minute of Atmee V1 in Behavior-driven Full-stack. What a credit costs in dollars depends on your plan; see Costs.
Plans
| Plan | Atmee V1 | Atmee V2 |
|---|---|---|
| Free | No API access | No API access |
| Starter | Yes | No |
| Growth, Pro, Scale and Enterprise | Yes | Yes, by invite for now |
Prices, monthly credits and limits for each plan are on the pricing page.
Choose the model of an avatar
The model is set when an avatar is created, and every session with that avatar uses it. To move an avatar to the other model, create a new one from the same portrait and voice.
Through the API, set engine in the avatar's manifest when you
create it. It is v1 when you leave it out:
{
"schemaVersion": 1,
"name": "Grand Valley Concierge",
"engine": "v2"
}Creating a V2 avatar before your account has V2 access answers 403 v2_access_required.
Next steps
Introduction
Atmee puts a talking video avatar in your product. It holds live, spoken conversations with your users, and you start each one from your own server.
Docs for AI agents
Every page of these docs, and the API's contract, as plain files an AI agent can read: an index, the whole docs in one file, a Markdown copy of each page, and the OpenAPI spec.