Compatible API guide

Anthropic-Compatible API Using the Claude Messages Format

Applications already structured around the Anthropic Messages API can call Infistar through the compatible endpoint while keeping the system, messages, and max_tokens request shape. Select the exact model ID from the live catalog rather than relying on a vendor display name or retired model name.

Send requests to POST /v1/messages. Include Authorization: Bearer YOUR_API_KEY, Content-Type: application/json, and anthropic-version: 2023-06-01. max_tokens is required and must be at least 1.

Messages API integration facts

POST /v1/messages

Anthropic Messages-compatible request endpoint

anthropic-version: 2023-06-01

Protocol version header required by the current public contract

Authorization: Bearer YOUR_API_KEY

Authentication with an Infistar API key

model

An exact Claude model ID from Models and Pricing

max_tokens

Required maximum output length, with a minimum value of 1

Migrate an Anthropic client

  1. Confirm that the client supports a custom API endpoint or base URL. A client locked to the vendor endpoint cannot connect by changing only the API key.
  2. Keep the Anthropic Messages role and content structure, and include the anthropic-version header.
  3. Choose a live model ID explicitly marked for the Claude or Anthropic endpoint.
  4. Test standard responses, streaming, system prompts, and tools separately because support depends on the selected model.

Minimal Messages API request

This example validates the protocol shape only. Replace MODEL_ID with a model visible to your API key that supports Messages API.

curl https://infistar.ai/v1/messages \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "MODEL_ID",
    "max_tokens": 256,
    "messages": [{"role": "user", "content": "Hello"}]
  }'

What compatibility does not imply

  • Messages compatibility does not automatically enable every Claude Beta header, experimental field, or vendor-specific feature.
  • Context, reasoning, tools, multimodal input, and streaming capabilities follow the current model facts and endpoint declaration.
  • For a 401 response, verify the API key. If a model is unavailable, refresh GET /v1/models and the live pricing page.