# Crouter

> Crouter is a unified AI API gateway that provides access to Claude, GPT, and Gemini models at ~90% lower cost than official APIs. We offer 11 production-ready models including Claude Sonnet 4.6 Thinking, GPT-5.2 Codex, and Gemini 3.1 Pro through a single OpenAI-compatible API. Designed for AI coding assistants (Claude Code, Cursor, Windsurf, Openclaw, Cline, Aider), SaaS applications, and developer tools. Pay-as-you-go pricing with no subscriptions, rate limits up to 70 req/min, and full streaming support.

## Product

- [Homepage](https://crouter.dev/): Overview of Crouter's unified AI API, pricing comparison showing 60-88% savings vs official APIs, and quick start guide for developers.
- [Pricing](https://crouter.dev/#pricing): Complete pricing table for all 11 models - Claude Haiku 4.5 ($0.12/$0.60), Sonnet 4.6 Thinking ($0.42/$2.10), Opus 4.6 Thinking ($0.84/$4.20), GPT-5.1/5.2/5.3 Codex ($0.50/$4.00), Gemini 3.1 Pro ($0.50/$3.00). All prices per 1M tokens with official API comparison.
- [Documentation](https://crouter.dev/docs): Complete API documentation including authentication, endpoints, request/response formats, error handling, rate limits, and integration guides for 6 AI coding tools.

## Integration Guides

- [Claude Code Integration](https://crouter.dev/docs#claude-code): Step-by-step guide to configure Claude Code CLI to use Crouter API with environment variables and base URL configuration.
- [Cursor Integration](https://crouter.dev/docs#cursor): Instructions for adding Crouter as custom model provider in Cursor IDE settings with API key and model configuration.
- [Windsurf Integration](https://crouter.dev/docs#windsurf): Setup guide for Windsurf IDE including API endpoint configuration and model selection.
- [Openclaw Integration](https://crouter.dev/docs#openclaw): Configuration steps for Openclaw to route requests through Crouter with custom provider settings.
- [Cline Integration](https://crouter.dev/docs#cline): Guide to configure Cline VS Code extension with Crouter API endpoint and authentication.
- [Aider Integration](https://crouter.dev/docs#aider): Setup instructions for Aider command-line tool including environment variables and model aliases.

## Models

- [Claude Models](https://crouter.dev/#claude-models): Three Claude models - Haiku 4.5 (88% off, $0.12/$0.60), Sonnet 4.6 Thinking (86% off, $0.42/$2.10), Opus 4.6 Thinking (83% off, $0.84/$4.20). All support prompt caching and 200K-1M context windows.
- [GPT Models](https://crouter.dev/#gpt-models): Six GPT models - GPT-5.1, GPT-5.1 Codex, GPT-5.1 Codex Max, GPT-5.2, GPT-5.2 Codex, GPT-5.3 Codex. All priced at $0.50/$4.00 (60-71% off official pricing) with 400K context.
- [Gemini Models](https://crouter.dev/#gemini-models): Two Gemini models - Gemini 3.1 Pro and Gemini 3.1 Pro High. Both $0.50/$3.00 (75% off official $2/$12 pricing) with 1M+ context and built-in reasoning.

## Resources

- [FAQ](https://crouter.dev/#faq): Common questions about API compatibility, billing, rate limits, model availability, streaming support, and migration from official APIs.
- [Contact](https://crouter.dev/contact): Support contact form for technical questions, billing inquiries, and partnership opportunities.

## Legal

- [Terms of Service](https://crouter.dev/terms): API usage terms, acceptable use policy, service level agreements, and liability limitations.
- [Privacy Policy](https://crouter.dev/privacy): Data handling practices, API request logging policies, and GDPR compliance information.

## Technical Details

- **API Format**: OpenAI-compatible REST API at api.crouter.dev/v1
- **Authentication**: Bearer token authentication via API keys from dashboard
- **Rate Limits**: 70 requests per minute per API key, upgradeable for enterprise
- **Supported Features**: Streaming responses, prompt caching (Claude only), function calling, JSON mode, temperature/top_p controls
- **Model Routing**: Automatic routing to Anthropic format (/v1/messages) for Claude models, OpenAI format (/v1/chat/completions) for GPT/Gemini
- **Billing**: Pay-as-you-go based on token usage, charged per 1M input/output tokens, no monthly minimums or subscriptions
- **Context Windows**: Haiku 200K, Sonnet/Opus 1M, GPT 400K, Gemini 1M+
- **Thinking Models**: Claude Sonnet/Opus 4.6 Thinking and Gemini 3.1 Pro include extended reasoning capabilities with thinking tokens billed as output

## Use Cases

- **AI Coding Assistants**: Primary use case - developers using Claude Code, Cursor, Windsurf, Openclaw, Hermes Agent, Cline, Aider with 60-88% cost savings
- **SaaS Applications**: Integrate Claude, GPT, or Gemini into products without managing multiple API keys or dealing with rate limits
- **Development Tools**: IDEs, code review tools, documentation generators, and CI/CD pipelines requiring LLM capabilities
- **Cost Optimization**: Teams spending $500+/month on official APIs can save $300-450/month by switching to Crouter
- **Multi-Model Testing**: Test same prompts across Claude, GPT, and Gemini through single API for model comparison and optimization

## Pricing Philosophy

Crouter achieves 60-88% discounts by using Quatarly's wholesale AI API access (50M credits for $15 = $0.30 per 1M credits). We maintain 25-40% margins while passing massive savings to developers. Official APIs charge premium prices; we provide the same models at near-cost pricing. No tricks, no rate limit games - just honest, transparent pricing for developers who want quality AI without the premium markup.