Which Mac runs which local AI model?

Pick a Mac to see which open models fit. Pick a model to see the cheapest Mac that runs it. Prototype - Will It Run Local (willitrunlocal.com).

Bandwidth ceiling = memory bandwidth ÷ bytes read per generated token. It is a theoretical upper bound, not a benchmark.

ModelFitWeights GiBKV GiBNeed GiBMeasured speed (public sources)
MacFitUS priceMemory usedMeasured speed (public sources)Buy

Data sources

Full methodology, assumptions and known gaps →