Migrating from OpenAI

Under maintenance. The managed Vinci API and Platform are under maintenance. These service-specific instructions are retained as reference for existing integrations when service is available. They do not promise current access or a return date. Downloadable models and direct/BYOK use of Vinci Code CLI are separate paths.

Vinci’s API is OpenAI-compatible, so migrating is usually a two-line change — the base URL and the API key — plus picking a Vinci model.

The change

  client = OpenAI(
-     base_url="https://api.openai.com/v1",
-     api_key=os.environ["OPENAI_API_KEY"],
+     base_url="https://vinci.getsimpledirect.com/api/v1",
+     api_key=os.environ["VINCI_API_KEY"],
  )
 
  resp = client.chat.completions.create(
-     model="gpt-4o",
+     model="forte",
      messages=[{"role": "user", "content": "Hello!"}],
  )

The supported chat-completions interface retains these request and response shapes. Check the reference below and test the features your integration uses.

What’s supported

• POST /chat/completions, streaming and non-streaming — see the reference.
• GET /models and GET /models/{model} — OpenAI-compatible model discovery.
• messages with system / user / assistant / tool roles, and content as a string or an array of parts.
• tools / tool_choice — native function calling with real tool_calls in the response. See Tool calling.
• stream, temperature, max_tokens.
• OpenAI-shaped responses (chat.completion, chat.completion.chunk) and error objects.

Differences to know

• Models are Vinci class ids — pin mezzo, forte, or fortissimo, or use auto for the account’s class routing, not gpt-*. Forte was the documented account default. Confirm current routing with the service when it is available; see models.
• The character layer is always applied. A system message is honored as additional guidance layered on top of — not replacing — Vinci’s voice.
• Not yet implemented: embeddings, the legacy /completions endpoint, logprobs, and n > 1. Tell us what you need.
• Zero data retention by default, with Canadian data residency for the data you store — see Privacy & data retention.

Keep your tooling

Clients that support the documented OpenAI-compatible chat interface can use its base URL, key and model settings. Compatibility depends on the client features used; test your integration when service is available. See Getting started for retained setup examples.