Discord · API reference · Configuration · Installation · Testing
Plexus sits in front of your LLM providers so clients can use one consistent API while you choose how requests are translated, routed, observed, and recovered. It supports OpenAI, Anthropic, Gemini, OpenAI-compatible providers, OAuth-backed subscriptions, and streamable HTTP MCP servers.
- Keep client code stable. Accept OpenAI Chat Completions and Responses, Anthropic Messages, Gemini native requests, embeddings, audio, images, streaming, and tool use.
- Route on your terms. Map aliases to one or more targets with
random,in_order,cost,performance,latency, ore2e_performanceselection. - Operate with evidence. Inspect request logs, tokens, cost, latency, live throughput, provider health, and per-key quotas from the dashboard.
- Stay resilient. Apply exponential cooldowns, fail over failed providers, detect stalled streams, and use vision fallthrough for non-vision targets.
ADMIN_KEY is required for the dashboard and management API. DATABASE_URL is optional and defaults to SQLite at ./data/plexus.db; use a PostgreSQL connection string for production.
docker run -p 4000:4000 \
-v plexus-data:/app/data \
-e ADMIN_KEY="your-admin-password" \
-e ENCRYPTION_KEY="your-generated-hex-key" \
ghcr.io/mcowger/plexus:latestDownload a pre-built binary from GitHub Releases:
# macOS Apple Silicon
curl -L https://github.com/mcowger/plexus/releases/latest/download/plexus-macos -o plexus
chmod +x plexus
ADMIN_KEY="your-admin-password" ./plexus
# Linux x64
curl -L https://github.com/mcowger/plexus/releases/latest/download/plexus-linux -o plexus
chmod +x plexus
ADMIN_KEY="your-admin-password" ./plexus# Windows x64
Invoke-WebRequest -Uri "https://github.com/mcowger/plexus/releases/latest/download/plexus.exe" -OutFile "plexus.exe"
$env:ADMIN_KEY = "your-admin-password"
$env:DATABASE_URL = "sqlite://./data/plexus.db"
.\plexus.exeThe binary is self-contained; database migrations and the web dashboard are embedded. See Installation for Docker Compose, Windows troubleshooting, source builds, and environment variables.
Open the dashboard at http://localhost:4000, then create/configure an API key and model alias. Send a request:
curl -X POST http://localhost:4000/v1/chat/completions \
-H "Authorization: Bearer sk-plexus-my-key" \
-H "Content-Type: application/json" \
-d '{"model": "fast", "messages": [{"role": "user", "content": "Hello!"}]}'OAuth providers are configured in the Admin UI. See Configuration: OAuth providers.
The dashboard is the operating view: configure providers and aliases, investigate every request, and watch usage and quota health.
| Dashboard | Providers |
|---|---|
![]() |
![]() |
| Request logs | Model aliases |
![]() |
![]() |
Protocol translation handles OpenAI, Anthropic, Gemini, and compatible formats in both directions, including streaming and tool use. MCP proxying isolates sessions for streamable HTTP MCP servers. Encryption at rest protects API keys, OAuth tokens, provider secrets, and MCP headers with AES-256-GCM when ENCRYPTION_KEY is set. See Configuration for the full operating model.
Pass a subcommand as the first argument to the binary or bun run src/index.ts:
rekeydecrypts sensitive fields with the currentENCRYPTION_KEYand re-encrypts them withNEW_ENCRYPTION_KEY.
ENCRYPTION_KEY="<current-key>" NEW_ENCRYPTION_KEY="<new-key>" ./plexus rekeybun run setup:hooks
bun run testbun test is intentionally blocked; use bun run test. See Testing.
MIT License — see LICENSE.



