Local-first agent · No cloud · No billing

Your AI employee. $0.00 forever.

Zero-Meter runs a fully autonomous AI agent on the machine you already own. Ollama under the hood, LFM2.5 in the chair — no tokens, no meter, no monthly bill.

macOS-first · 3-minute setup · no account, no card
zero-meter — zsh
Runs on Ollama Hermes LFM2.5 macOS
Why Zero-Meter

Everything a subscription does.
Nothing a subscription costs.

We removed the meter entirely. What's left is a tireless local agent with no usage caps — because there is no usage bill.

No meter

Unlimited tokens because there is no token bill. Run your agent around the clock, on every task you can throw at it — the counter on the wall reads $0.00 and always will.

∞ inferences / day

Fully private

Every prompt, file, and tool call stays inside your machine. No API calls to the cloud, no telemetry on your work — the network cable is optional, not required.

0 bytes leave your machine

Swarm-ready

One agent is an employee; five are a department. Spawn parallel workers that divide a job, share results, and finish together — still for $0.00.

N agents · 1 invoice: none
0tok/s
on-device throughput
0K context
window, all local
0GB RAM
minimum footprint
$0.00
what you'll ever pay
How it works

Three commands to a permanent hire.

No account. No card. No meter to watch. If you can open a terminal, you can staff an AI employee today.

01

Install the stack

One line pulls Ollama, the quantized LFM2.5 model, and the Zero-Meter runtime. Everything lands on your disk — nothing phones home.

$ zero-meter install
02

Point it at work

Give your agent a role and scoped access to files, shell, and tools. It plans, executes, and reports back — entirely in local RAM.

$ zero-meter init --role ops
03

Let it run forever

Queue tasks, schedule runs, or hand it a standing order. Zero-Meter keeps it alive day and night — and the bill never arrives.

$ zero-meter run --forever
Pricing

One plan. No fine print.

Free forever no trial · no expiry
$0.00/ month

The full product, on your hardware, for as long as you want it. Upgrades included, forever.

  • Unlimited local inference — no token meter, ever
  • LFM2.5 quantized for 8GB+ machines (185 tok/s on Apple Silicon)
  • Swarm mode: parallel agents at no extra cost
  • 100% private — works fully offline
  • All future releases, $0.00 each
Get Zero-Meter free
brew install zero-meter · that's the whole checkout
FAQ

Fair questions.

There isn't one — because there's nothing to resell. Zero-Meter runs on hardware you already own, so there's no GPU bill for us to pass along. We build the runtime; your machine does the computing. The price is $0.00 today, next year, and after that.
Any Apple Silicon Mac with 8GB of unified memory is the comfortable minimum — that's where the 185 tok/s number comes from. A Linux box with a recent 8GB+ GPU works too. Bigger RAM buys you longer context and larger swarms, not a bigger bill.
Yes — verifiably. Every inference runs through Ollama on localhost, and the agent has no cloud endpoints compiled in. You can unplug the network cable mid-task and Zero-Meter keeps working; that's the test we design against.