Third-party AI API relay service

Ship Faster With
AI Proxy Relay

KRouter is a third-party intermediary that relays AI model API requests through rotating proxies, manages quota distribution, and provides a unified OpenAI-compatible endpoint. Build in hours, not weeks.

JoinClaude CodeOpenAI CodexOpenCodeOpen ClawFactory DroidCursor+7 more

Live routing animation

terminalClaude Code
terminalCodex CLI
terminalCline
terminalCursor
hubKRouter
OpenAI
Anthropic
Google AI
iFlow AI

Core Value

KRouter sits between your tools and upstream AI providers as a third-party intermediary.

lan

Third-party relay for AI APIs

KRouter sits between your client tools and upstream AI providers. It is not the official endpoint of OpenAI, Anthropic, Google, or GitHub.

sync_alt

Proxy rotation and quota distribution

Requests can be relayed through rotating proxy pools, multiple accounts, and fallback routes so quota is spread more evenly and sessions are less fragile.

contract

One base URL for coding tools

Most supported tools only need one base URL, one API key, and one chosen model path to begin working with KRouter.

Features

Everything you need to manage AI API routing in one place.

model_training

Unified API surface

Expose a single OpenAI-compatible entry such as `https://KRouter.net/v1` to clients that support custom providers.

route

Quota-aware routing behavior

The service is positioned around usage control, quota balancing, and keeping sessions alive when one route becomes unstable.

shield_person

Explicit third-party disclosure

The landing makes it clear that KRouter is a third-party intermediary and that upstream provider policies still matter.

description

CLI-first documentation

A dedicated Docs page collects the setup instructions for Claude Code, Codex, OpenCode, Open Claw, Cursor, Cline, Continue, and more.

deployed_code

Proxy and MITM paths where needed

For tools like Antigravity, Copilot, and Kiro, the docs explain the MITM-oriented flow instead of pretending every tool works the same way.

translate

Bilingual communication

Core positioning and setup content are written in English and Vietnamese so the product can onboard a wider local audience cleanly.

How KRouter Works

Requests flow from your coding tools through our relay layer to the best available upstream provider.

settings_input_component01

Choose the right connection path

Use the public KRouter endpoint for OpenAI-compatible tools, or use MITM-based setup for tools that do not natively support custom local providers.

tune02

Configure API key and model mapping

Each CLI guide mirrors the current dashboard setup flow: base URL, API key, model selection, and where to place the configuration file.

monitoring03

Route, rotate, and monitor usage

KRouter then acts as an intermediary layer for model APIs, rotating paths and helping manage quota pressure across multiple upstream channels.

Connect Your Tools

Point your CLI tools to the KRouter endpoint. One base URL, one API key, one model path — that's all most tools need.

1

Set Base URL

Use https://krouter.net/v1

2

Add API Key

Copy your KRouter API key from the dashboard

3

Choose a Model

Select the model path you want to route

menu_bookView All CLI Guides
settings.json
Claude Code — settings.jsonClick to copy
{
  "env": {
    "ANTHROPIC_BASE_URL": "https://krouter.net/v1",
    "ANTHROPIC_AUTH_TOKEN": "YOUR_KROUTER_API_KEY"
  }
}
terminal
Claude Code
Managed config
terminal
OpenAI Codex CLI
Managed config
terminal
OpenCode
Managed config
terminal
Open Claw
Managed config
query_statsUsage Checker

Check Your API Key Usage

Enter your API key below to view your current usage, limits, and expiration status.

vpn_key
gavel

Important Notice

Please read and understand the following before using KRouter.

priority_high

KRouter is a third-party intermediary service, not an official AI model vendor.

priority_high

Although KRouter is not an official provider, all models are called through legitimate OAuth credentials and rotating API keys sourced directly from the original providers — ensuring every response is authentic and unmodified from the upstream source.

priority_high

Proxy rotation, account rotation, and fallback behavior are used to help distribute quota and reduce interruption risk.

priority_high

Users are still responsible for complying with upstream provider terms, local law, and their own operational security requirements.