# Concurred API

> Concurred provides a single API to access 8 top AI models: Claude (Anthropic), GPT (OpenAI),
> Grok (xAI), Gemini (Google), DeepSeek, Kimi (Moonshot), Mistral, and Llama (Meta).
> One API key for all models. OpenAI-compatible gateway. Supports single-model chat,
> side-by-side battle mode, multi-agent fight mode with voting and leaderboards,
> multi-instance mode (4x same model, like Grok Heavy), autonomous web search with citations,
> image/vision analysis across 4 models, and AI-powered fashion image generation.
>
> Base URL: https://concurred.ai
> Authentication: Bearer token (Authorization: Bearer KEY) or X-API-Key header

## Getting Started

- [Overview](https://concurred.ai/docs/api): Platform overview, available models table, key features summary
- [Quick Start](https://concurred.ai/docs/api/quickstart): Get running in minutes with curl, Python, and JavaScript examples for all modes
- [Authentication](https://concurred.ai/docs/api/authentication): Bearer token and X-API-Key authentication methods, endpoint-specific auth table
- [Public Keys](https://concurred.ai/docs/api/public-keys): Browser-safe `ck_pub_` keys for frontend apps (Vite, Vue, Three.js, Flutter). Origin-bound, spend-capped, rate-limited. Safety model: origin is a speed bump, spend cap is the ceiling.

## API Reference

- [Gateway API](https://concurred.ai/docs/api/gateway): OpenAI-compatible endpoint POST /api/v1/chat/completions — drop-in replacement for OpenAI SDK, supports streaming, temperature, max_tokens, vision (multimodal messages)
- [Chat API](https://concurred.ai/docs/api/chat): Unified endpoint POST /api/chat — chat mode (1 model), battle mode (2 models side-by-side), fight mode (2-8 models with voting/leaderboard), multi-instance mode (4x same model), web search with citations, image_url for vision, conversation history
- [Fashion API](https://concurred.ai/docs/api/fashion): 10 AI fashion generation models — virtual try-on, product-to-model, face-to-model, model-create, model-variation, model-swap, reframe, background-change, background-remove, image-to-video. Async submit/poll or SSE subscribe pattern.
- [Vision API](https://concurred.ai/docs/api/vision): Standalone Grok vision endpoint POST /api/v1/vision — image analysis via URL or base64, supports PNG/JPG/GIF/WEBP
- [Models](https://concurred.ai/docs/api/models): All 8 model IDs, full alias list per model, vision capability matrix

## Resources

- [Error Codes](https://concurred.ai/docs/api/errors): All error codes with HTTP status codes, example error response bodies, troubleshooting solutions
- [Rate Limits](https://concurred.ai/docs/api/rate-limits): Per-endpoint rate limits, rate limit response headers, exponential backoff example

## Optional

- [Full Documentation](https://concurred.ai/llms-full.txt): All documentation pages concatenated as plain text — fetch once for complete API knowledge
- [OpenAPI Spec](https://concurred.ai/openapi.json): Machine-readable OpenAPI 3.0 specification for all endpoints