Can a Mac Studio M3 Ultra, 256GB run local LLMs?

Yes — 38 of the 39 open models on this site fit in its 192 GB of usable memory, the strongest being GLM-5.3-Flash. Whether that saves you money is a different question, and the answer is usually no.

Price$7,099 at launch — discontinued

Memory256 GB, about 192 GB of it addressable by the GPU at 819 GB/s

Best model it runsGLM-5.3-Flash — Sonnet-class, 22 tok/s

Pay-back against the APIPays back in 781 years at 500k tokens a day

Run the numbers on this machine

What it runs

Model	Speed	Class	Memory
GLM-5.3-FlashUD-Q4_K_M	22 tok/s	Sonnet-class	190 GB
Qwen3.8 Flash NextQ4_K_M	50 tok/s	Sonnet-class	121 GB
DeepSeek V4-FlashUD-Q4_K_M	34 tok/s	Sonnet-class	155 GB
Qwen3.8 27BQ4_K_M	33 tok/s	Sonnet-class	19 GB
Inkling SmallUD-Q4_K_M	29 tok/s	Haiku-class	164 GB
Ling 3.0 flashQ4_K_M	35 tok/s	Haiku-class	82 GB
MiniMax M2.7Q4_K_M	17 tok/s	Haiku-class	149 GB
Qwen3.6 27BQ4_K_M	32 tok/s	Haiku-class	19 GB
Qwen3.6 35B-A3BQ4_K_M	98 tok/s	Haiku-class	23 GB
Muse Glimmer 30BQ4_K_M	34 tok/s	Haiku-class	18 GB
Gemma 4 26B-A4BQ4_K_M	72 tok/s	Haiku-class	18 GB
Qwen3.5 122B-A10BUD-Q4_K_M	35 tok/s	Haiku-class	79 GB

26 more fit; the calculator lists them all.

The specifics

Chip: M3 Ultra — 32-core CPU / 80-core GPU
Memory bandwidth: 819 GB/s
Usable by the GPU: 192 GB — macOS lets the GPU wire roughly 75% of unified memory by default (llama.cpp discussion #2182: 66.7% at 32 GiB or less, 75% above). Raise it with `sudo sysctl iogpu.wired_limit_mb`. Memory is treated as GB throughout, which is slightly conservative.
Power under load: 270 W (published) — Apple's published 'max' figure for this chip (support article 102027), a CPU+GPU stress number; inference typically draws less.
Availability: Discontinued 2026-08-25. Launch price; Apple no longer sells it new, so treat as a refurbished/used reference. 256GB required the 32-core chip.
Sources: source 1, source 2, source 3