Reference

OpenAI SDK (TypeScript)

Use the official OpenAI Node.js SDK with anymize as the base URL.

1npm install openai

OpenAI SDK (Python)

Use the official OpenAI Python SDK with anymize as the base URL.

1pip install openai

Base URLs

Different base URLs are used depending on the feature.

Supported formats

The API accepts JSON (application/json) and multipart/form-data for file uploads.

Anthropic SDK (TypeScript)

Official @anthropic-ai/sdk. Point the base URL at /api; the client appends /v1/messages. apiKey is sent as x-api-key, which anymize accepts.

1npm install @anthropic-ai/sdk

Anthropic SDK (Python)

anthropic.Anthropic with base_url on /api. api_key sends x-api-key (works). auth_token sends Bearer, same as Claude Code.

1pip install anthropic

Vercel AI SDK (OpenAI provider)

createOpenAI with baseURL .../api/v1/llm talks Chat Completions. generateText, streamText, tools: same as api.openai.com, plus anonymization.

1import { createOpenAI } from "@ai-sdk/openai"
2import { generateText } from "ai"
3
4const openai = createOpenAI({
5  apiKey: "YOUR_API_KEY",
6  baseURL: "https://app.anymize.ai/api/v1/llm",
7})
8
9const { text } = await generateText({
10  model: openai.chat("fountain-1.0"),
11  prompt: "Hello",
12})

Default limits

Each API key has default limits per minute and per day.

Rate limit headers

Every response includes headers with current limits and usage.

Rate limit error

When exceeded, you'll get a 429 error with a Retry-After header.

{
  "error": {
    "message": "Rate limit exceeded. Please retry after 30 seconds.",
    "type": "rate_limit_error",
    "code": "rate_limit_exceeded"
  }
}

Retry strategy

Wait for the time specified in the Retry-After header before retrying.

1# Check rate limit headers in the response
2curl -i -X POST https://app.anymize.ai/api/v1/llm/chat/completions \
3  -H "Authorization: Bearer YOUR_API_KEY" \
4  -H "Content-Type: application/json" \
5  -d '{
6    "model": "fountain-1.0",
7    "messages": [{"role": "user", "content": "Hello!"}]
8  }'
9
10# Response headers include:
11# X-RateLimit-Limit-Requests: 60
12# Retry-After: 30

Credit-based limits

In addition to rate limits, credit-based limits apply depending on your plan.

Higher limits

Need higher limits? Contact us for an enterprise plan.

HTTP status codes

All API endpoints use standard HTTP status codes.

Error format

Errors are returned as JSON with an error object.

{
  "error": {
    "message": "Missing required field: messages",
    "type": "invalid_request_error",
    "code": "missing_messages",
    "param": "messages"
  }
}

Error types

Each error has a type that describes its category.

Error codes

Specific error codes for common problems.

Error examples

{
  "error": {
    "message": "Invalid API key",
    "type": "auth_error",
    "code": "invalid_api_key"
  }
}

{
  "error": {
    "message": "Model 'gpt-nonexistent' not found",
    "type": "invalid_request_error",
    "code": "model_not_found",
    "param": "model"
  }
}

{
  "error": {
    "message": "Rate limit exceeded. Please retry after 30 seconds.",
    "type": "rate_limit_error",
    "code": "rate_limit_exceeded"
  }
}

{
  "error": {
    "message": "The model is currently overloaded. Please try again later.",
    "type": "api_error",
    "code": "service_unavailable"
  }
}

Best practices

Recommendations for robust error handling.

• •Always check the HTTP status code before parsing the response.
• •Log errors with error.code and error.message for debugging.
• •Implement retry logic with exponential backoff for 429 and 5xx errors.
• •Show the end user a helpful error message, not the raw API error.

114 maintained integration operations within 467 routes and 673 total endpoints.