Blog · Site Guide · September 20, 2026 · 5 min read
Welcome to RunAIHow: What This Site Is and How to Use It
.webp)
RunAIHow is a free, privacy-friendly calculator that answers two questions every person interested in local AI eventually asks: “Can my hardware run this AI model?” and “Is it cheaper to run it locally or pay for a cloud API?” There is nothing to install, no account to create, and no data leaves your browser — every calculation happens on your device.
The three panels
- Hardware & Model Compatibility — pick your GPU (or type a custom one), choose a quantization level and context length, and see instantly which of the 90 models in the library fit your VRAM, with an estimated generation speed for each.
- Token Cost — paste a typical prompt, set your expected output length and request volume, and see what that usage would cost per month across every cloud model in the library.
- Local vs Cloud Crossover — combine your hardware price, electricity rate, and workload into a ten-year cost chart that marks the exact month your own GPU becomes cheaper than the cloud.
A one-minute walkthrough
Start in the Hardware panel: search for your graphics card — its VRAM, memory bandwidth, and power draw fill in automatically. Editing any value switches you to a custom GPU; the “Number of GPUs” field handles multi-card setups. Then scroll the compatibility table: green means an excellent fit, yellow means it runs but tight, red means it will not fit at that quantization and context. Drag the context slider to see why long conversations change the answer.
Curious about the money side? Describe your typical request in the Token Cost panel — the site estimates tokens from your text — and set how many requests a day you would make. Finally, the crossover chart puts your GPU and that cloud bill on the same timeline and shows the break-even point.
Built for beginners, honest for experts
All estimates use simplified but transparent math (bandwidth-bound speed, linear KV-cache growth, FP16-based quantization ratios), and the formulas are shown in tooltips so you can check them. Your settings are remembered on your device between visits, the site works on phones, and a blog covers the deeper topics — starting with our Ollama install guide. When you are ready, scroll up and pick your GPU — the answer to “can I run it?” is one glance away.