InferWatch

InferWatch · Press kit

InferWatch press kit.

Everything you need to write about InferWatch: descriptions, a fact sheet, the app icon and full-size screenshots. Each file is a direct link; no zip, no sign-up.

InferWatch app icon

Description

About InferWatch

InferWatch is a native iPhone monitor for self-hosted LLM servers. It shows live tok/s, KV-cache pressure and queue depth for vLLM, installed models for Ollama, and gateway health and token stats for LiteLLM, and sends a local notification when a rig backs up, stalls or drops offline.

People who run their own inference keep checking the same few numbers by SSH or a Grafana tab. InferWatch connects straight to their servers, with no account and no backend, and puts those numbers on the iPhone: a KV-cache ring, decode tok/s, time to first token, and running and waiting requests. One server is free. A one-time Pro unlock adds unlimited rigs, breach alerts, history charts and a Home Screen widget.

InferWatch was called TokenWatch before launch.

Fact sheet

The facts.

App Store name
InferWatch: LLM Monitor
Developer
Forgehaven Labs LLC
Platform
iPhone, iOS 17 or later
Category
Developer Tools, Utilities
Price
Free download for one server. InferWatch Pro: one-time $14.99
Privacy
Data Not Collected. No account, no analytics
Works with
vLLM, Ollama, LiteLLM, Prometheus /metrics with vLLM metric names
Status
Coming soon to the App Store
Website
forgehavenlabs.com/inferwatch
Contact
[email protected]

Downloads

Screenshots are the App Store set for iPhone 6.9 inch displays. Preview them in the gallery.

The studio

About Forgehaven Labs

Forgehaven Labs LLC is a small software studio that builds privacy-first iPhone apps and tools. Its monitor apps, InferWatch, FireGlance and FinDeck, read your own servers directly: no account, no backend, no tracking.

For press questions, review access or interviews, write to [email protected] with "Press" in the subject.

InferWatch is an independent app and is not affiliated with or endorsed by the vLLM, Ollama, or LiteLLM projects. Their names are used only to describe compatibility.