KushRouter API Documentation

Welcome to the KushRouter API docs.

• Main unified endpoint: POST /api/v1/messages
• OpenAI-compatible streaming: POST /api/openai/v1/chat/completions
• Anthropic-compatible streaming: POST /api/anthropic/v1/messages

Quick start

• Get an API key from your dashboard: kushrouter.com/dashboard/api-keys (opens in a new tab)
• Try our browser Playground: kushrouter.com/dashboard/playground (opens in a new tab)
• Use your favorite SDK and point it to our base URLs

Try it now

• Open Playground with a prefilled prompt for the unified endpoint

  • Model: gpt-4o-mini
  • Prompt: “Write a creative haiku about AI routers”
  • Launch Playground (opens in a new tab)

• Try streaming with Anthropic (prefilled)

  • Model: claude-3-5-sonnet-v2@anthropic
  • Prompt: “Summarize: Why use a unified LLM router?”
  • Launch Streaming (opens in a new tab)

Base URLs

• Production base URL: https://api.kushrouter.com
• Unified: https://api.kushrouter.com/api/v1/messages
• OpenAI: https://api.kushrouter.com/api/openai/v1/
• Anthropic: https://api.kushrouter.com/api/anthropic/v1/

What’s covered here

• Endpoints for streaming chat across providers
• Request schemas and validation behaviors
• CLI integration for OpenAI Codex and Anthropic Claude Code
• Pricing, model aliases, and credit consumption rules
• Usage analytics exports and dashboards
• File uploads, batch processing, tokenize helper, and MCP server integrations
• Troubleshooting guidance for common issues

If you encounter an error response, the message will be descriptive and aligned with the request validators described in the Schemas section.