Cheapest Mac to run DeepSeek V4 Pro
1600B parameters (49B active) · quality index 52 · coding 70 · every Apple Silicon Mac ever sold, compared. It is a mixture-of-experts model: all 1600B parameters must sit in memory, but only 49B compute per token, which is why it is faster than dense models of similar size.
No single Mac can hold DeepSeek V4 Pro at practical quantizations — it needs ≈1039GB resident. Running it locally means clustering multiple machines (see the cluster planner on the main site), or using a cloud API at about $0.87/1M output tokens.
Run it
ollama run deepseek-v4 pulls the default (≈Q4) build once you have Ollama installed.
Or skip the hardware
Cloud APIs serve DeepSeek V4 Pro at about $0.87 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 →