Connect your AI tools, apps, and agents to one OpenAI-compatible endpoint. 8Router handles provider routing, quota tracking, circuit breaker, and fallback so your tools stay connected when keys expire or models go down.
$ npm install -g 8router
$ 8router
# API live at http://localhost:8080/v1
# Dashboard at http://localhost:8080/8router/dashboardRoute chat requests across connected providers with health-aware fallback.
ActiveUse one /v1 endpoint with tools that support custom OpenAI base URLs.
ActivePick the right provider based on quota, health, latency, and route mode.
ActiveTrack key usage, limits, cooldowns, and exhausted providers in one place.
ActiveUse aliases like 8router/cheap, 8router/fast, and 8router/coding.
BetaRoute tool-capable requests only to models that support function calling.
BetaFallback before streaming starts, then keep response chunks stable.
BetaVector embeddings endpoint with multi-provider fallback.
Coming SoonImage generation endpoint with provider routing.
Coming Soon
Best available route based on health, quota, latency, and fallback rules.
Chooses the cheapest healthy model available.
Chooses the lowest-latency healthy model.
Routes to the strongest available model for reasoning-heavy tasks.
Optimized for coding, debugging, tool use, and long context tasks.
Local-only route. Never falls back to cloud providers.
npm install -g 8routerAdd API keys or local providers from the dashboard.
Set base URL in Cursor, Cline, or OpenWebUI:
http://localhost:8080/v1Use 8router/auto, 8router/cheap, 8router/coding, or 8router/local.
Track quota, key health, fallback paths, and latency from the dashboard.
12 providers with dedicated adapters for request normalization, response parsing, and health checks.
30+ models tracked for vision, tools, streaming, embeddings, context length, and cost.
Score-based model selection using cost, speed, quality, and latency data.
Per-model pricing with input/output token costs, currency, and free quota tracking.
p50/p95/p99 latency tracking per provider with automatic health monitoring.
Automatic provider cooldown after 5 consecutive failures. 3-minute reset window.
Track key status: healthy, rate_limited, exhausted, invalid, cooldown. Auto-skip bad keys.
Per-provider quota tracking across 5h, daily, weekly, and monthly periods.
API keys masked in all responses, logs, dashboard, and error messages.
Error messages sanitized to remove sensitive data before logging or returning.
Drop-in replacement for /v1/chat/completions, /v1/models, /v1/embeddings.
Ready-to-copy configs for Cursor, Cline, Continue, OpenWebUI, Claude Code, and Codex.
Base URL: http://localhost:8080/v1
API Key: ***
Model: 8router/autonpm install -g 8router8routerBase URL: http://localhost:8080/v1
Model: 8router/auto