bitHuman

Real-time AI avatars with lip-synced video. Stream audio in, get a talking face out at 25 FPS — running on your hardware or ours.

Full product docs at docs.bithuman.ai.

Where to start

You want to…	Start at	What you get
Try it in 5 minutes	docs.bithuman.ai/getting-started/quickstart	`pip install bithuman` and run a working example end-to-end.
Run runnable examples	bithuman-sdk-public	Python, Swift, CLI, and REST examples plus integration recipes (LiveKit, Next.js, Java, Gradio, offline Mac).
Embed the on-device Swift SDK	bithuman-sdk-public	SwiftPM binary package — `import bitHumanKit`. Mac M3+ / iPad M4+ / iPhone 16 Pro+.
Install the Mac CLI	homebrew-bithuman	`brew install bithuman-product/bithuman/bithuman` — voice + video chat from Terminal, 100% on-device.
Install the Python SDK	PyPI: bithuman	`pip install bithuman`. Runs Essence avatars on any CPU (laptop, server, Raspberry Pi).
Use the REST API	docs.bithuman.ai/api-reference	No local runtime — manage avatars and sessions over HTTPS from any language.
See a reference app	bithuman-apps	Working Mac / iPad / iPhone / Flutter apps that consume the SDK. Clone, build, talk to an avatar.

	Essence (start here)	Expression
What it is	Pre-built `.imx` file generated from a photo or video	Any face image, animated at runtime
Hardware	Any CPU — laptop, server, Raspberry Pi	NVIDIA GPU or Apple Silicon M3+
Best for	Voice agents, kiosks, 24/7 displays, getting started	Apps where users supply their own face

Releases — Swift SDK on GitHub Releases, Python SDK on PyPI, CLI on the Homebrew tap.
Issues & questions — file in bithuman-sdk-public/issues (covers both Swift and Python SDKs) or join the Discord.
Security reports — email security@bithuman.ai. See each repo's SECURITY.md for scope.
Status, updates, blog — bithuman.ai · @bithumanAI.

Each repo carries its own license — most are MIT or source-available with a build-and-run grant. See the LICENSE file in each repo.