Skip to main content
Corvex Token Factory exposes OpenAI-compatible inference endpoints, Anthropic-compatible Messages endpoints, a public model catalog, and account-scoped usage reporting. The endpoint schemas below are generated from the public OpenAPI 3.1 specification in this repository.

​
Base URL

OpenAI SDKs normally append /v1 in their configured base URL. Anthropic SDKs use the host-level URL shown above.

​
Authentication

Inference and usage requests require a Token Factory API key. Send Authorization: Bearer sk-corvex-... to OpenAI-compatible and usage endpoints. Send x-api-key: sk-corvex-... to Anthropic-compatible endpoints. GET /v1/models and its /openai/v1/models alias are public. A valid key is optional on these two list operations. Create and revoke keys in the Token Factory dashboard. See Authentication for storage and rotation guidance.

​
Endpoint groups

Chat Completions

POST /v1/chat/completions supports OpenAI-compatible streaming and non-streaming chat requests.

Responses

POST /v1/responses supports the documented stateless subset of the OpenAI Responses API.

Messages

POST /v1/messages and POST /v1/messages/count_tokens provide the Anthropic-compatible surface used by the Anthropic SDK and Claude Code.

Models

GET /v1/models returns the public live catalog, capabilities, limits, and published rates.

Legacy text completions

POST /v1/completions accepts models that advertise text-completion support. Use Chat Completions for the current catalog.

Embeddings

POST /v1/embeddings is defined, but no embedding model is currently deployed.

Usage

GET /api/v1/usage returns usage scoped to the authenticated account.

​
Current models

Call GET /v1/models for the live catalog. Currently available:
  • zai-org/GLM-5.3
  • deepseek-ai/DeepSeek-V4-Flash-0731
Use the model ID exactly as returned. The model catalog compares context windows, image input, capabilities, and pricing.
Last modified on September 22, 2026