Keep the official OpenAI client and Chat Completions request format. Change the
client’s base URL and API key, then choose a model from the Token Factory
catalog. Model IDs and feature support differ from OpenAI’s hosted models.
Prerequisites
- A Token Factory API key. See Authentication.
- Python with the
openai package, or Node.js with the openai package.
- A model ID from the live catalog.
Set the key in your shell:
The highlighted constructor fields route the client to Token Factory. Requests
use the standard Bearer header generated by the OpenAI SDK.
Run the repository examples
A non-2xx response raises the SDK’s API error type with the structured Token
Factory error body. Configure timeouts and retries through the SDK as you would
for any remote inference service.
Model IDs
Token Factory model IDs use namespace/model format. List the public catalog
before hard-coding an ID:
Do not add a corvex/ prefix. The model catalog documents
input types, context windows, capabilities, and rates.
- Quickstart — test the endpoint with
curl.
- Errors — handle authentication, validation, rate-limit,
and server failures.
- API reference — review the supported OpenAI-compatible
request and response fields.
Last modified on September 15, 2026