Using the API

The server speaks the OpenAI API format, so most AI tools and SDKs work with it once their base URL is changed.

Three rules apply to every request:

• Stream long answers with "stream": true. The connection is dropped after 100 seconds with nothing sent.
• Send a User-Agent header. Python’s default Python-urllib agent is blocked with error 1010, while curl, browsers and the OpenAI SDK are accepted.
• Tool calling is switched off: tools are removed from every request.

Examples

With curl:

curl -N https://ai.sfuracerbot.ca/v1/chat/completions \
  -H 'Content-Type: application/json' \
  -d '{"model":"LFM2.5-8B-A1B","stream":true,
       "messages":[{"role":"user","content":"Explain PID control in two sentences."}]}'

With the OpenAI Python SDK:

from openai import OpenAI

client = OpenAI(base_url="https://ai.sfuracerbot.ca/v1", api_key="none")
stream = client.chat.completions.create(
    model="LFM2.5-8B-A1B",
    messages=[{"role": "user", "content": "Give me three robot names."}],
    stream=True,
)
for chunk in stream:
    print(chunk.choices[0].delta.content or "", end="", flush=True)

Working when: the answer prints a few words at a time.

Per-request controls