Quickstart

Change one base URL. Keep your prompts, tools, and SDKs.

Where to point today

Environment	base_url	Notes
Local Docker (Rust edge)	http://localhost:8081/v1	Supported MVP contract; bootstrap key sk-at-dev
Documented production host	https://api.withohm.dev/v1	Hostname reserved for the Ohm edge; until the edge is cut over, treat chat calls as not yet public — use local or your operator deploy

Keys use the legacy prefix sk-at-… (product brand is Ohm). Headers use the X-AT-* family until rename.

Design-partner path

1. Request a key (email partners@withohm.dev, or operator issues POST /v1/admin/tenants).
2. Point your OpenAI SDK at the edge base_url above.
3. After a week, check GET /v1/usage for cache_hit_ratio and arbitrage_gross_usd.
4. Accept Terms and DPA (terms_ack / dpa_ack) when using web context.

Ten lines of Python (local)

from openai import OpenAI

client = OpenAI(
    base_url="http://localhost:8081/v1",
    api_key="sk-at-dev",
)

r = client.chat.completions.create(
    model="gpt-4o-mini",
    messages=[{"role": "user", "content": "Hello"}],
)
print(r.choices[0].message.content)

Identical second call → Redis cache hit (header x-at-cache: HIT).

Compliance

Web context requires purpose + acks. See Legal and Privacy.