Designed for wide screens · min 1024px → open the mobile site
A compatibility engine for local AI

I want to turn text into text⌄ — on this machine⌄ .

Every model that can do this — scored against your exact machine.

scroll

There are — public models speed-tested by the community. — can do what you asked.

motherboard — — — —
or pick a reference rig
—
—
text → text
intent · text inference
won't run
runs, tight
comfortable
headroom
quantization
— fit → — comfortable
—
—
—weights + KV cache
0
16k
32k
64k
128k
—
tokens / sec
live preview text → text

Three models that actually fit your machine and your intent.

can-i-run-it engine verified benchmarks Rust CLI MCP server hardware telemetry model index community configs

This isn't a database.
It's the operating system for local AI compatibility.

Benchmarks are community-submitted and re-verified on reference rigs.
Read how scoring works →
How scoring works
  1. measured — the median of many community runs on your exact rig and quantization.
  2. reported — one or two community runs on the same rig, shown honestly.
  3. extrapolated — scaled by memory bandwidth from the closest tested rig.
  4. formula — a published VRAM rule when no measurement exists yet.
  5. no data yet — when we have nothing, we say so instead of inventing a number.