Cheapest Mac to run Qwen3 235B-A22B
235B parameters (22B active) · quality index 40 · coding 45 · every Apple Silicon Mac ever sold, compared. It is a mixture-of-experts model: all 235B parameters must sit in memory, but only 22B compute per token, which is why it is faster than dense models of similar size.
The cheapest Mac that runs Qwen3 235B-A22B comfortably is a used Mac Studio M3 Ultra 32c/80c 256GB (2025) at about $5,887 EST. on the used market — running Q5_K_M quantization at roughly 26 tok/s with up to 76K tokens of context.
Every Mac that runs it, by used price
| Machine | Used price | Runs at | Est. speed | Max context |
|---|---|---|---|---|
| Mac Studio M3 Ultra 32c/80c 256GB (2025) | $5,887 EST. | Q5_K_M | 26 tok/s | 76K tokens |
| Mac Studio M3 Ultra 32c/80c 512GB (2025) | $7,457 EST. | Q8_0 | 17 tok/s | 128K tokens |
Run it
ollama run qwen3:235b pulls the default (≈Q4) build once you have Ollama installed.
Or skip the hardware
Cloud APIs serve Qwen3 235B-A22B at about $0.6 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 →