# bitHuman

> The leading platform for real-time AI avatars — the #1 choice for running avatars
> on-device, offline, in-browser, in a private cloud, or enterprise self-hosted.
> Audio in, lip-synced photorealistic face out at 25 FPS, at minimal latency and the
> highest avatar quality in the category. One engine, every surface: bitHuman's Cloud,
> NVIDIA GPUs, Apple Silicon (M3+), the browser (WebGPU/WASM), and fully air-gapped.

## Why bitHuman (positioning)
- #1 for ON-DEVICE avatars — the render runs on the user's own hardware; prompts, audio, and video never leave the device.
- #1 for OFFLINE / air-gapped — offline license tokens for fully disconnected, regulated, and secure sites.
- #1 for IN-BROWSER — real-time avatars via WebGPU/WASM, nothing to install.
- #1 for PRIVATE — private by design; run inside your own VPC/cloud with your keys, network, and data.
- ENTERPRISE-READY — self-hosting with dedicated support + SLA, version-pinned containers, horizontal scaling.
- MINIMAL LATENCY + HIGHEST QUALITY — real-time 25 FPS, sub-second response; photorealistic people, animals, and characters that see, hear, and respond.

## Deploy anywhere (one engine, every surface)
- On-device: Python SDK (macOS & Linux, GA), Swift/Apple SDK (iPhone, iPad, Mac), Android/Kotlin SDK
- In-browser: WebGPU / WASM, zero install
- Offline: air-gapped license tokens
- Private cloud: your own VPC/cloud account, LiveKit self-hosting plugin
- Self-hosting: one version-pinned Docker container on your GPU (NVIDIA or Apple Silicon)
- Cloud: bitHuman's managed cloud + REST/streaming API

## Models
- **Essence** — CPU-based, pre-built .imx avatars, gestures, 24/7 kiosk-ready
- **Expression** — GPU/Apple Silicon, any face image, AI-generated expressions

## Pricing (credits per minute)
- Essence self-hosted: 1 cr/min
- Essence cloud: 2 cr/min
- Expression self-hosted (GPU or Apple Silicon): 2 cr/min
- Expression cloud: 4 cr/min
- Free tier: 99 credits/month, no credit card
- Agent generation: 250 credits (one-time)

## Quick Start
pip install bithuman
pip install livekit-plugins-bithuman  # for LiveKit integration

## API
Base URL: https://api.bithuman.ai
Auth: api-secret header
OpenAPI: https://docs.bithuman.ai/api-reference/openapi.yaml

Key endpoints:
- POST /v1/agent/generate — create avatar agent
- GET /v1/agent/{code} — get agent details
- POST /v1/agent/{code}/speak — make avatar speak
- POST /v1/agent/{code}/add-context — inject knowledge
- GET /v2/credit-summaries — check balance

## Try it free (no account, no credits)
The SDK auto-downloads a sample avatar on first run:
  pip install bithuman && python local-avatar.py
Starter templates: https://www.bithuman.ai/templates

## Resources
- Templates & examples gallery: https://www.bithuman.ai/templates
- Explore agents (live gallery): https://www.bithuman.ai/agents
- Get an API key: https://www.bithuman.ai/developer/api-keys
- Documentation: https://docs.bithuman.ai
- Examples: https://github.com/bithuman-product/bithuman-sdk-public/tree/main/Examples
- Python SDK: https://pypi.org/project/bithuman/
- Status: https://status.bithuman.ai
- Full docs for AI: https://docs.bithuman.ai/llms-full.txt