Quick start

Updated 2026-09-23

Four steps take you from sign-up to your first call: sign up and verify your email, create an API key, pick a model in the catalog, and send a request to a compatible endpoint.

1. Sign up and verify your email

You are signed in right after registering. Creating API keys, topping up and redeeming codes all require a verified email. The verification link is valid for 24 hours and works once. If the instance offers trial credit, it is granted as gift credit once your email is verified.

2. Create an API key

Open the Keys page in the console and give the key a name. The full key starts with key_ and is shown once right after creation, so save it immediately. To see it again later you need to enter your login password; see API keys.

3. Pick a model

Find a model in the catalog. The "Model ID" on its detail page is the value for the model field. The detail page also lists capabilities, input limits and prices in the instance currency.

4. Make your first call

Three compatible APIs are served on this site's domain:

Call the chat endpoint with curl:

curl https://omnimodel.me/v1/chat/completions \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "<model-id>", "messages": [{"role": "user", "content": "Hello"}]}'

With the OpenAI Python SDK:

from openai import OpenAI

client = OpenAI(base_url="https://omnimodel.me/v1", api_key="<key>")
reply = client.chat.completions.create(
    model="<model-id>",
    messages=[{"role": "user", "content": "Hello"}],
)
print(reply.choices[0].message.content)

Defaults worth knowing

• Streaming: add "stream": true. The OpenAI format ends with data: [DONE].
• Without max_tokens (or max_completion_tokens) the request is treated as 4096. max_tokens is required in the Anthropic format.
• Keys are accepted in request headers only; a key in the URL is rejected.
• GET /v1/models lists only the models your key may call.
• Insufficient balance returns billing.insufficient_credit (HTTP 402); top up and retry. See common errors for more.