Teale ∙ manifesto | for OpenClaw | for Hermes | for Developers

Teale is an OpenAI-compatible API backed by the world's idle laptops.

Install in one line. Swap the base URL. Point at teale-auto and let the compiler pick the right model for every request.

Install the CLI

curl -fsSL https://teale.com/install.sh | sh

macOS and Linux via install.sh. Linux desktop: download the x86_64 tarball. Windows: download Teale.exe — the CLI ships bundled in the app. Early builds are unsigned; bypass SmartScreen on first launch.

Use any OpenAI SDK

from openai import OpenAI

client = OpenAI(
    base_url="https://gateway.teale.com/v1",
    api_key="tok_...",
)

response = client.chat.completions.create(
    model="meta-llama/llama-3.1-8b-instruct",
    messages=[{"role": "user", "content": "Hello"}],
)

Or let teale-auto pick the model

teale-auto is a compiler that routes each request to the right model — small for simple prompts, large for reasoning, with fallbacks across the network.

response = client.chat.completions.create(
    model="teale-auto",
    messages=[{"role": "user", "content": "Hello"}],
)
# → teale-auto routes to Hermes Llama 8B today; more models as supply scales.

Models

Hermes Llama 8B is the primary served model today. More canonical OpenRouter-slug models are added as the network scales.

Pricing per model: full docs →