PAI Router

Model infrastructure you can operate with confidence.

Choose a model and keep one stable interface in your product. Router handles execution while preserving the facts you need to understand every request.

API surface

Published model

claude-fable-5

POST /v1/responsesStreaming supported
curl https://router.omnitools.ai/v1/responses \
  -H "Authorization: Bearer $PAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-fable-5",
    "input": "Explain what makes a reliable model router.",
    "max_output_tokens": 160
  }'

Supported fields are explicit, and unsupported parameters return a clear response.

From model choice to observable request.

Router keeps execution details behind one product-facing model interface.

  1. 01

    Create an API key

    Issue a dedicated key for the service or environment making requests.

  2. 02

    Choose a model

    Use a stable PAI model ID across the supported API surfaces.

  3. 03

    Send the request

    Router selects a compatible route and records usage against the model price.

  4. 04

    Review the result

    Follow one Request ID from response through usage and charge.

One model ID, resilient routing.

Automatic routing prefers native protocol support, then weighs recent availability and internal cost.

Model
claude-fable-5
Protocol
Responses / Chat Completions / Messages

A stable boundary for your product.

Your application keeps one model interface while PAI manages routing behind it.

Security and data boundaries

Run your first request.

Open Console to choose a model, create an API key and review the result.

Open PAI Console