Documentation

Everything you need to integrate Izzi API into your stack.

Getting Started

Install Izzi API in seconds with our auto-configuration tool:

npx izziapi

This command auto-detects your installed tools (Claude Code, Cursor, Cline, Windsurf) and configures them to use the Izzi API endpoint.

Authentication

All API requests require an API key. Include it in the header:

Authorization: Bearer izzi-xxxxxxxxxxxx

Get your API key from the dashboard after signing up.

API Endpoints

POST/v1/chat/completions

OpenAI-compatible chat completions endpoint.

curl https://api.izziapi.com/v1/chat/completions \
  -H "Authorization: Bearer izzi-xxxxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-sonnet-4-20250514",
    "messages": [
      {"role": "user", "content": "Hello!"}
    ]
  }'

POST/v1/messages

Anthropic-compatible messages endpoint.

POST/v1/responses

OpenAI Responses API (simplified). See details below.

POST/v1/messages/count_tokens

Pre-count input tokens before making a request.

GET/v1/models

List all available models with pricing info.

GET/v1/models/:id

Lookup a specific model by ID. Supports aliases.

Also available at: /openai/v1/* and /api/v1/*

Responses API

Simplified interface compatible with n8n, LangChain, and other tools using OpenAI's Responses API format.

curl https://api.izziapi.com/v1/responses \
  -H "Authorization: Bearer izzi-xxxxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-sonnet-4-20250514",
    "input": "Explain quantum computing in simple terms",
    "instructions": "You are a helpful teacher"
  }'

input — string or array of messages

instructions — system prompt (optional)

stream — enable streaming (optional)

Count Tokens

Pre-count input tokens before making a request — useful for cost estimation.

curl https://api.izziapi.com/v1/messages/count_tokens \
  -H "Authorization: Bearer izzi-xxxxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-sonnet-4-20250514",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

For Anthropic models, uses native token counting. Other models use local estimation (~4 chars/token).

Extended Thinking

Enable Claude's Extended Thinking for complex reasoning tasks:

curl https://api.izziapi.com/v1/messages \
  -H "x-api-key: izzi-xxxxx" \
  -H "Content-Type: application/json" \
  -H "anthropic-version: 2023-06-01" \
  -d '{
    "model": "claude-sonnet-4-20250514",
    "max_tokens": 16000,
    "thinking": {
      "type": "enabled",
      "budget_tokens": 10000
    },
    "messages": [{"role": "user", "content": "Solve this step by step..."}]
  }'

Supported models: Claude Opus 4, Claude Sonnet 4. The thinking parameter is transparently forwarded to Anthropic.

Rate Limits

Every response includes rate limit headers:

x-ratelimit-tier: pro
x-ratelimit-limit-requests: 60
x-ratelimit-remaining-requests: 47
x-ratelimit-reset-requests: 2026-03-30T00:20:00Z
x-concurrent-limit: 3
x-concurrent-remaining: 2
Retry-After: 60  (only on 429)

Plan	RPM	Concurrent	Daily
Free	10	1	100
Starter	30	2	1,000
Pro	60	3	2,000
Max	90	6	5,000
Ultra	120	8	10,000

Claude Code Setup

Configure Claude Code to use Izzi API:

# Set environment variables
export ANTHROPIC_BASE_URL=https://api.izziapi.com
export ANTHROPIC_API_KEY=izzi-xxxxx

# Start Claude Code
claude

Cursor Setup

Open Cursor → Settings → Models
Add https://api.izziapi.com/v1 as the OpenAI Base URL
Enter your Izzi API key
Select your preferred model

Supported Models

claude-opus-4-20250514claude-sonnet-4-20250514claude-3-5-haiku-20241022gpt-5gpt-4.1gpt-4.1-minigemini-2.5-progemini-2.5-flashgrok-3step-3.5-flashglm-4.5-air

Error Codes

All errors follow a consistent format:

{
  "error": {
    "type": "rate_limit_error",
    "message": "Rate limit exceeded. Try again in 60 seconds."
  }
}

HTTP	Type	Description
400	invalid_request_error	Malformed request or missing fields
401	authentication_error	Invalid or missing API key
402	insufficient_balance	Not enough credits
403	ip_blocked	IP blocked after too many failed auth attempts
429	rate_limit_error	RPM or concurrent limit exceeded
429	quota_exceeded	Daily request quota exceeded
402	budget_exceeded	Monthly budget limit reached
404	not_found	Model or resource not found
502	upstream_error	Upstream AI provider returned an error
502	connection_error	Failed to reach upstream provider
500	server_error	Internal server error