Get started

Quickstart

Make your first request in under a minute. Use the managed hosted API, or run the open weights yourself — both speak the OpenAI Chat Completions API.

Flywheel models are niche AI employees — each one fine-tuned for a single vertical (auto shops, law firms, gyms, the trades). Every model speaks the OpenAI Chat Completions API, so you can reach it with any OpenAI SDK by changing one line.

Call the hosted API

The fastest path. Mint an API key on the Account page (keys look like fw_live_…), export it, and send a request. The model is the niche slug — see the catalog for the full list.

curl https://gyld.dev/api/v1/chat/completions \
  -H "Authorization: Bearer $FLYWHEEL_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "fitness",
    "messages": [
      { "role": "user", "content": "Beginner full-body workout?" }
    ]
  }'

from openai import OpenAI

client = OpenAI(base_url="https://gyld.dev/api/v1", api_key="fw_live_…")

resp = client.chat.completions.create(
    model="fitness",
    messages=[{"role": "user", "content": "Beginner full-body workout?"}],
)
print(resp.choices[0].message.content)

You’ll get back the standard OpenAI completion object — the assistant message is at choices[0].message.content. From here, see the full Chat completions reference.

Tip.Already have OpenAI code? Point its base_url at https://gyld.dev/api/v1, swap in your fw_live_ key, and set the model to a niche slug. Nothing else changes — details on OpenAI compatibility.

Run it yourself

Prefer to own the runtime? The weights are public on Hugging Face under flywheel-ai/<niche> (Apache-2.0). Serve them and you get the same OpenAI-compatible endpoint on your own machine — nothing leaves your network.

shell

pip install vllm
vllm serve flywheel-ai/fitness --served-model-name fitness
# OpenAI-compatible endpoint now at http://localhost:8000/v1

That’s a full local /v1 endpoint. Point any OpenAI SDK at it exactly as above. For laptop-friendly GGUF builds and hardware sizing, see Self-hosting.

Next steps

Authentication — how API keys work, scopes, and rotation.
Chat completions — every request and response field.
The model family — pick the right niche, sizes, and quantizations.
Errors & rate limits — status codes and how to handle them.

Authentication →