meterless is prepaid GPU compute with the temperament of a phone plan — the name is the policy: a stated monthly allowance, storage that stays put (and stays priced) when you stop, and your first result minutes after signup.
Launching in small groups · No card needed to join the list
Every line on the left is the kind of clause — or documented complaint — GPU renters live with today. Every line on the right is meterless's standing commitment, written into our terms.
Quotes are from public reviews of other GPU platforms. We keep the full sources — ask us.
One page, four plans, plain English. Pay by card. No quotes, no sales calls, no cloud console.
One-click templates, pre-configured and ready — or a blank GPU with SSH if that's your thing.
Your first render, minutes after your card. Models and outputs live on your persistent disk.
The fastest path we offer: from idea to running GPU without opening a browser. Launch, pause, resume and check your hours from Claude, Cursor, or any MCP client. Tokens issued from your dashboard — read-only by default, control scope when you say so. Purchases stay human, always.
> claude mcp add meterless https://api.meterless.io/mcpWe measure this and publish the real number: target median under 5 minutes to first result.
The dashboard is deliberately boring: how many hours you have, whether your GPU is running, and one button to open your workspace. The interesting part is what it never shows — a running total in dollars.
illustration of the meterless dashboard — design shown, not a screenshot
Run the big open video models the way they were released — without squeezing them through quantized workarounds.
Multi-LoRA pipelines, upscalers and control stacks loaded together — and the end of “CUDA out of memory”.
Serve 70B-class models with Ollama on hours you already own — and via MCP, your agent can even pause your GPU on the way out.
LoRA and fine-tuning runs on professional-class memory, with your datasets on a disk that's always there.
A meterless plan is one complete package. Elsewhere, the typical metered cloud sells the pieces separately — each on its own meter, each on its own clock.
The hardware is comparable across the industry. The terms are the product.
Pick the hours, not the hardware. Every plan runs the same professional-class capacity — storage, environments and the make-good included.
Evenings & weekends creating
The daily driver for builders & creators
For heavy weeks & serious workloads
For work that truly needs 96 GB
Prices in USD. VAT or local sales tax is added at checkout where your country requires it — shown clearly before you pay, never after. Workload figures are typical — your mileage varies with models & settings. Need different hours or a specific card? Tell us when you join — plans get added when enough of you ask, not because a catalogue exists.
meterless is deliberately built lean and heavily automated — so the promises don't depend on anyone's goodwill on a given day. The commitments are written down, and the systems enforce them: capacity failures credit your plan automatically, storage is never held hostage, and nothing bills you silently — ever.
Support is async and honest: stated response targets, no tiers, no outsourced script-readers. Every ticket is read by the people who run the platform, and the status page tells you what we know when we know it.
The make-good policy, in full → · Security & your data → · Status page →
Your GPU pauses and we ask what you'd like to do — top up, upgrade, or wait for renewal. Your files stay on your disk either way. Nothing ever silently bills you.
Our systems detect it and credit the time back to your plan automatically. If we ever take payment and can't provision your plan at all, you get a full refund. Both commitments are in the terms, in plain English.
No. One-click environments open in your browser, pre-configured. If you do want SSH and a raw machine, that's there too.
No — and we want this crystal clear. Tools like Claude Code come pre-installed, but they run on your own account and API keys: Anthropic (or any other AI vendor) bills you directly for what you use, and meterless never marks it up or touches it. If you'd rather pay nothing per token, that's exactly what the local route is for: run open models on your own GPU with Ollama — those use only the hours you already own.
meterless runs on professional data-center capacity from vetted infrastructure partners, packaged and guaranteed by us. Your contract, support and make-good are all with meterless — one accountable party.
On encrypted persistent volumes in professional data centers. Deleting your account deletes your volumes — stated plainly on our Security page.
[First cohorts open — LAUNCH WINDOW TBC]. We onboard in small groups so every early customer gets a flawless start — the list is the queue.
Yes. Cancel in one click, effective at the end of your paid month — no calls, no retention scripts. Your storage stays available through a stated grace period so you can export or come back.
Base plan hours don't roll over — that's part of what keeps flat pricing possible. Purchased top-ups do roll into your next month.
Yes. Your studio is a real machine with SSH, so Cursor and VS Code connect straight in with their remote features — your editor and your API keys on your laptop, our GPU doing the heavy lifting. Claude Code is also ready inside every studio if you prefer working in the terminal, using your own key, billed by Anthropic.
Not at launch. Payments are processed securely by Stripe, with cards from day one. Stablecoin payment is on the roadmap and switches on the moment Stripe supports it for businesses in our region — the price will be identical and we will never see or hold your keys. If paying in crypto matters to you, tell us when you join the waitlist and we will let you know the day it goes live.
We're opening in small cohorts so every early customer gets a flawless start. Two ways in:
A handful of honest build updates, then your invitation when your cohort opens. No spam — ever.
Fully refundable any time before launch. Founding members get first access and lock their launch price for 12 months.
Reserve my spot — $25Refunded automatically if we don't launch. No subscription starts until you say so.
Email only. No spam, no card needed for the list.