Cheapest Mac to run DeepSeek-V3 / R1
671B parameters (37B active) · quality index 37 · coding 49 · every Apple Silicon Mac ever sold, compared. It is a mixture-of-experts model: all 671B parameters must sit in memory, but only 37B compute per token, which is why it is faster than dense models of similar size.
No single Mac can hold DeepSeek-V3 / R1 at practical quantizations — it needs ≈436GB resident. Running it locally means clustering multiple machines (see the cluster planner on the main site), or using a cloud API at about $2.5/1M output tokens.
Run it
ollama run deepseek-r1:671b pulls the default (≈Q4) build once you have Ollama installed.
Or skip the hardware
Cloud APIs serve DeepSeek-V3 / R1 at about $2.5 per million output tokens. The main site's break-even solver computes the daily usage where owning a Mac becomes cheaper than renting.
Open the interactive guide — speed simulator, TCO, all 75 machines →