Observed arrival · 2026-10-06
EndoRouter: a local-first model router
An open-source gateway that routes OpenAI-compatible chat requests according to privacy labels and provenance.
○Open
⊠Login
$Paid
†Ads
✦Pretty
●Pro
◎Niche
◉Human
⚑Risk
ƒJS
- For
- Developers routing private prompts through OpenAI-compatible clients
- Worth noticing
- Setup requires Python 3.10 or newer and an already-running local model server; unsupported request types are refused rather than forwarded.
Field notes
At startup, the CLI looks for an already-running local model server, checks the program behind its port, and adds cloud providers only when their keys are present. The page says the gateway handles text chat completions, including streaming, and refuses images, audio, embeddings, and the Responses API rather than forwarding them. Its benchmark uses two recording servers and publishes cases, methods, results, and a limited threat model.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
○OpenPublic substance visible
✦PrettyNotable craft visible
●ProPolished or operationally mature
◎NicheUnusually specific use
One card from the complete issue