API documentation

Create a key in your account under API keys. It is shown once, so copy it then. Send it as Authorization: Bearer, as the OpenAI clients do, or as x-api-key, as the Anthropic clients do. A key belongs to the account it was created under, and every call made with it is that account's usage. Revoking a key stops it working within seconds.

Pick a model from the pricing page, then call it.

curl https://sven.swedishembedded.com/v1/chat/completions \
  -H "Authorization: Bearer $SVEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "MODEL", "messages": [{"role": "user", "content": "Hello"}]}'

import os
from openai import OpenAI

client = OpenAI(base_url="https://sven.swedishembedded.com/v1", api_key=os.environ["SVEN_API_KEY"])
reply = client.chat.completions.create(
    model="MODEL",
    messages=[{"role": "user", "content": "Hello"}],
)
print(reply.choices[0].message.content)

Set stream to true and the reply arrives as server-sent events, exactly as the OpenAI client expects. If the connection drops, the reply still runs to its end and you are charged for what it produced; send the same request again with the same Idempotency-Key header to receive the whole answer without being charged again (while it is still running, the retry is answered 409: try again shortly).

You pay for what a call delivered, at the prices on the pricing page: per token for chat, per input token for embeddings, per image for images. Before a call starts, the most it could cost is set aside from your account's credit, so a call your credit cannot cover is refused up front rather than failing halfway. When it finishes you are charged for what it actually used and the rest is available again.

• 400 The request is malformed, or sets a parameter the model does not accept (the error names it), or a conversation longer than the model can serve.
• 401 No key, or a key that is unknown, expired or revoked.
• 402 The account's credit, the key's spending limit or the workspace's monthly budget does not cover what this call could cost. The error code says which; retrying will not help until it does.
• 403 This key is restricted to other models.
• 404 No such model. A model that exists but has no price is also absent: it is not for sale.
• 409 A call with this Idempotency-Key is still running (retry shortly), or its stored answer has expired.
• 413 The request body is past the size ceiling.
• 429 Too many calls at once or per minute, or the model is busy. Retry after the Retry-After header's seconds.
• 500 Something failed inside the service. The response says nothing about internals. Retrying is safe but may fail again.
• 503 The model is temporarily unavailable. Retry shortly.

The full description of every route is available as OpenAPI, for generating a client.