Free
$0
Watch one rig, forever.
- One server, live
- tok/s, TTFT, queue depth, KV-cache
- vLLM · Ollama · LiteLLM · Prometheus
- Demo mode, no server required
- Unlimited rigs & fleet view
- Push-on-OOM alerts
Forgehaven Labs · TokenWatch · iOS · for people who run their own inference
Watch your LLM rigs from your phone.
TokenWatch is a native iPhone monitor for your self-hosted inference. Live tok/s, KV-cache pressure, and queue depth for your vLLM, Ollama, and LiteLLM servers, and a push the moment one OOMs or drops offline.
Coming soon to the App Store · iOS 17+ · one-time unlock, no subscription, no account
The at-a-glance layer
Is my rig healthy right now?
The question every self-hoster asks and can't answer from their phone. Stop SSH-ing in or leaving a Grafana tab open on a laptop. TokenWatch connects straight to your servers and shows the numbers that actually tell you.
What it does
The telemetry that matters, glanceable.
KV-cache pressure
A ring you can read across the room. Watch the cache fill before it forces evictions and tanks your throughput.
Live tok/s & TTFT
Real decode throughput and time-to-first-token, derived from the Prometheus counters your server already exposes.
- the reason to install it
Push-on-OOM
Get pinged the instant a rig OOMs, backs up its queue, or drops offline. One alert per incident, no re-nagging.
Whole-fleet view
Every server at once in one health grid. Fleet tok/s, rigs online, and a tap into any rig's detail.
History & widgets
Trend tok/s and KV-cache over time, and glance at rig health from your Home Screen.
Nothing leaves your device
No account, no backend, no analytics. TokenWatch talks only to the servers you add. Your tokens and URLs stay yours.
On your phone
Built to be read in two seconds.
Who it's for
For the people who run the servers.
Homelab & LocalLLaMA
You run vLLM or Ollama on a box in the closet. TokenWatch is the glance-at-your-phone view you've been faking with a browser tab.
Small teams on a shared rig
One GPU box, several people hammering it. Watch queue depth and KV-cache so you know when it's saturated, from anywhere.
LiteLLM gateway operators
Running LiteLLM in front of your models? TokenWatch reads its health, and its token metrics when you enable the Prometheus callback.
Anyone who got burned by a silent OOM
You found out hours later that the server died mid-run. Push-on-OOM makes that the last time.
Pricing
Free to watch. Pay once for the fleet.
No subscription. No ads. The single-server live monitor is free forever; a one-time unlock watches everything.
TokenWatch Pro
$14.99
One-time. Lifetime.
- Everything in Free
- Unlimited rigs & whole-fleet view
- Push-on-OOM alerts
- Historical charts
- Home-screen widgets
- Family Sharing included
One-time purchase · no subscription
FAQ
Questions, answered.
What servers does it work with?
Anything exposing Prometheus metrics (vLLM, llama.cpp, TGI), plus the Ollama API and LiteLLM gateways. Point it at your metrics endpoint and it maps the fields automatically, including the vLLM v0-to-v1 KV-cache metric rename.
Do I need to open my server to the internet?
No. Reach it however you already do: a LAN IP on the same Wi-Fi, or a VPN like Tailscale. TokenWatch requests the local-network permission so it can talk to homelab boxes directly.
Is it really private?
Yes. There is no TokenWatch backend and no account. The app talks only to your servers; bearer tokens live in the iOS Keychain and never leave the device. No analytics, no ads, no tracking. Its App Store privacy label is Data Not Collected.
Subscription?
No. TokenWatch Pro is a one-time purchase that unlocks everything, forever. The single-server monitor is free.
Can I try it before I have a server set up?
Yes. Demo mode loads a full sample fleet so you can explore every screen with no server at all.
Coming soon
Know your rigs are healthy, from anywhere.
TokenWatch is heading to the App Store. Drop a line and I'll tell you the day it lands.
iOS 17+ · one-time purchase · no subscription