Tools

Calculators for the decisions that come before a purchase. Each one shows the formula it uses, states what it cannot know, and says where its numbers came from.

Runs in your browser. Nothing you type is uploaded, logged or sent anywhere, because there is nowhere for it to go: these pages are static files and the arithmetic happens on your machine. Same principle as the rest of the site.

Where each tool's numbers come from

Runs on your numbers

Nothing is seeded. This tool computes from what you type in, so it cannot be more accurate than the figure you give it.

Researched

Seeded with published specifications and cited sources rather than results measured here. Every seeded figure links to where it came from.

• Model fit calculator

  Will this model run on my GPU, and at what context length?

  Weights, KV cache and runtime overhead against the memory you have, with the formula shown and the file-size question answered along the way.

  Researched

• Power cost calculator

  What does this machine cost me a year?

  Idle and load draw to annual kilowatt hours, annual cost, cost per hour of inference and energy cost per million tokens.

  Researched

• Local vs cloud break-even

  Is running AI locally actually cheaper than the API?

  Hardware price and electricity against API pricing, with the months to break even and the honest verdict when the answer is never.

  Researched

• Tokens per second simulator

  What does 8 tokens per second actually feel like?

  Stream text at any generation speed, side by side with a second speed, and find out what a benchmark number means before you buy hardware for it.

  Runs on your numbers

• NAS capacity and power

  How much space do I actually get, and what does it cost to run?

  Drive count and layout to usable capacity in both TB and TiB, failures survived, rebuild window, and the standing power bill nobody else calculates.

  Researched

• Noise combiner

  Two machines at 38 and 41 dB(A). How loud is that together?

  Decibels add logarithmically, so 38 and 41 make about 43 rather than 79. Combine any number of sources at any distance and see what actually changes.

  Researched

• UPS sizing calculator

  How many VA do I need, and how long will it actually last?

  Measured draw to a VA and watt rating with the power-factor conversion shown, plus a runtime ceiling that says plainly why it is a ceiling.

  Researched

What is coming

The queue is public so the gaps are visible. Two tools are held back rather than shipped on estimates, because both of them are only worth anything once there is measured data to put in them.

• Tokens per watt comparator: pick machines and models from the bench, get tokens per watt and cost per million tokens against current API pricing. Waiting on published runs, and it is the page the bench exists to make possible
• Home lab cost index: every machine on the bench ranked by three-year total cost of ownership, including measured electricity. Also waiting on runs

Neither is a build problem. Both need more than an inference row: tokens per watt wants measured watts, and the cost index wants electricity. The first published run is speed and memory, not those tools.