InferWeave

Connect

Point any OpenAI-compatible client at the weave: the gateway endpoint, the /v1 surface, how to obtain a key, and copyable examples.

Fixture data (no live surface yet).

Endpoint

Point your client at: https://gw.example

  • GET /v1/models — List the models published through the gateway.
  • POST /v1/chat/completions — Create a chat completion (OpenAI-compatible).
  • POST /v1/responses — Create a response (OpenAI-compatible).
  • POST /v1/embeddings — Create an embedding (OpenAI-compatible).

Published models

  • inferweave/demo-7b

API key

Key issuance is not available in this environment yet; the instructions below are the standing procedure.

  1. Request a scoped API key from your gateway operator; keys are issued per caller and can be revoked.
  2. Keep the key in an environment variable such as INFERWEAVE_API_KEY; never commit or log it.
  3. Send it as a bearer token: Authorization: Bearer <your key>.

Examples

List the published models

curl https://gw.example/v1/models \
  -H "Authorization: Bearer $INFERWEAVE_API_KEY"

Create a chat completion

curl https://gw.example/v1/chat/completions \
  -H "Authorization: Bearer $INFERWEAVE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "inferweave/demo-7b",
    "messages": [
      { "role": "user", "content": "Say hello." }
    ]
  }'