Meterless

What is Meterless?

Meterless is a local-first AI suite built around Nightshift, which turns a Mac or DGX Spark into an AI hotspot for your apps. When you step away, it races model attempts on queued jobs and keeps the best. Free Gaia, Relay, and Swarms apps add memory, workflows, and agent teams.

Top Features:

• AI hotspot: serves text, image, video, and speech models through one OpenAI-compatible endpoint.
• Overnight racing: tries hundreds of attempts, clusters them, and judges results overnight.
• Compounding methods: each night distills what worked into reusable methods and skills.

Use Cases:

• Business work: queue strategy memos, OKRs, or RFP answers and review drafts by morning.
• Creative exploration: generate many directions for video, music, or image projects at once.
• Private serving: connect chat, voice, and transcription apps to models on your machine.

Who Can Use Meterless?

• Knowledge workers: hand off long research briefs and documents to run while away.
• DGX Spark owners: put idle workstation hours to use with unattended, verified AI work.
• Private teams: keep files and models local without per token cloud bills.

Pricing

• Web edition (free): run a first job in your browser with nothing to install.
• Mac edition ($199 one-time): preorder a perpetual license with 1.x updates, shipping Q4 2026.
• DGX edition (contact sales): runs on NVIDIA DGX Spark, with release details shared on request.

Pros and Cons

Pros:

• No token meter: local models cost only electricity, about five cents per deliverable.
• Many models: works with over 190 local and cloud models across media types.
• Checked output: each kept result includes a record of validation and judging steps.

Cons:

• Not shipped yet: the Mac and DGX editions are still preorder or awaiting release.
• Hardware limits: the Mac edition needs an M5 chip or later with 16 GB.
• Busy lineup: several apps, engines, and editions make the product hard to grasp.

FAQs:

1) Do files leave my computer?

Not with local models; cloud providers see data only when you add keys.

2) Is there a free version?

Yes, the web edition and the three desktop apps are free.

3) Which hardware is supported?

Apple Silicon Macs with M5 chips or later, plus NVIDIA DGX Spark workstations.

4) Can my apps connect?

Yes, any client that speaks the OpenAI API connects through one local URL.

5) What happens overnight?

Models race attempts, cluster similar ones, verify outputs, and keep the best work.