Can I run it? Check any AI model against your GPU or Mac ($0 markup)
Pick your GPU or Mac (or type your VRAM) and see which of the 377 open models in our catalog fit — and, for each, all three honest costs: run it locally ($0 compute + electricity), rent an equivalent GPU ($0 markup, as of 2026-07-02), or pay per-token via your own API key (as of 2026-06-29).
Pick your hardware — see what fits and what it really costs
The most-asked question in local AI: "what can MY machine run, and what does it cost?" Choose a card or Mac below (or enter any VRAM number in the app) and we rank the 377 notable models in our catalog by whether they fit your memory — using the same VRAM math (weights + KV cache at 16k context + runtime overhead) as the rest of the site.
For every model that fits you get all three honest costs side by side, and the ones that don't fit are separated out with the cheapest way to run them anyway (a rented GPU or your own API key).
How many models fit each card or Mac
Used NVIDIA RTX 3090 24GB (24 GB VRAM): 281 of 377 models fit.
NVIDIA RTX 4090 24GB (24 GB VRAM): 281 of 377 models fit.
Used NVIDIA RTX 4090 24GB (24 GB VRAM): 281 of 377 models fit.
NVIDIA RTX 5090 32GB (32 GB VRAM): 298 of 377 models fit.
Apple MacBook Air M3 16GB (refurb) (16 GB unified): 243 of 377 models fit.
Apple Mac mini M4 Pro 24GB (24 GB unified): 277 of 377 models fit.
A short email of real AI price moves, straight from the daily log — no hype. We're collecting the list now; the first issue goes out when it opens. Unsubscribe with one click.