Open-model inference
Open models.One API.
Run the open-weights models everyone is talking about, served from GPUs we operate ourselves. OpenAI-compatible, billed per token, paid in USDC.
From prompt to production. On open weights.
Our own GPUs
Every model runs on hardware we operate. No reselling someone else's endpoint, no surprise rate limits from a provider behind us.
Trending, then live
We follow what is rising on Hugging Face and bring the best open-weights models online as soon as they are ready to serve.
Pay in USDC
Top up on Base or Robinhood Chain. Credit lands after the transfer confirms and never expires.
Drop-in compatible
Point the OpenAI SDK at our base URL. Chat completions, streaming, tool calls and reasoning work the way your code expects.
Quickstart
Change one line.
Keep the SDK you already use. Swap the base URL and the key, pick a model, and you are serving open weights.
- OpenAI-compatible /v1/chat/completions
- Server-sent event streaming
- Tool calling and reasoning output
- Per-request usage in every response
from openai import OpenAI
client = OpenAI(
base_url="https://openmodel.club/v1",
api_key="om_live_...",
)
resp = client.chat.completions.create(
model="MODEL_ID", # any id from GET /v1/models
messages=[{"role": "user", "content": "Hello!"}],
)
print(resp.choices[0].message.content)
Built for builders
Infrastructure you can reason about.
Plain per-token prices, usage you can audit to the request, and an API that behaves like the one you already know.
Priced per model
Input, cached input and output tokens each have a published price per million.
Billed to the millionth
Each request is charged from the server's own token counts, rounded up to $0.000001.
Balance never overdrawn
A request reserves its worst case first, so parallel calls cannot spend the same credit.
Keys you control
Create and revoke keys from the console. Every key's spend is visible per model.
Questions, answered.
Which models do you serve?
Open-weights models only, chosen from what is trending on Hugging Face and served on our own GPUs. The live list is always at /models and from GET /v1/models.
Do I need to change my code?
No. Any OpenAI SDK or HTTP client works: set the base URL to https://openmodel.club/v1 and use your OpenModel key.
How do I pay?
Send USDC on Base or Robinhood Chain from the console. Your balance is credited once the transfer confirms. There are no cards and no subscriptions.
What happens when my balance runs out?
Requests are refused with 402 out_of_credit until you top up. You are never charged more than your balance.
Can I ask for a model?
Yes. Tell us which Hugging Face repo you want served and we will weigh it for the next deployment.
Start building on open models.
Sign up, claim your free credit and make your first call in a minute.