What is the biggest model the Apple M1 Ultra can run?

gpt-oss 120B, which needs about 67 GB, on the 128 GB version at about 97 tok/s.

What is the biggest model the Apple M2 Ultra can run?

Qwen3 235B-A22B, which needs about 144 GB, on the 192 GB version at about 21 tok/s.

Apple M1 Ultra vs Apple M2 Ultra for local AI

The Apple M2 Ultra fits more: up to Qwen3 235B-A22B, against gpt-oss 120B on the Apple M1 Ultra. They generate at about the same speed, since their memory bandwidth is close (800 and 800 GB/s).

The numbers that matter

Memory and bandwidth

Memory decides which models fit. Bandwidth decides how fast they generate, because every token reads the model's active weights once.

	Apple M1 Ultra	Apple M2 Ultra
Memory	64 or 128 GB unified memory	64 or 128 or 192 GB unified memory
Usable for models	About 96 GB of 128 GB	About 144 GB of 192 GB
Bandwidth	800 GB/s	800 GB/s
Biggest model that fits	gpt-oss 120B	Qwen3 235B-A22B
Best everyday pick	gpt-oss 120B, ~97 tok/s	Qwen3 235B-A22B, ~21 tok/s

Model by model

Every model, side by side

Estimated generation speed at the largest memory size each was sold in (128 GB and 192 GB).

Model	Apple M1 Ultra	Apple M2 Ultra
Qwen3 4B About 4 GB	~180 tok/s Runs great	~180 tok/s Runs great
Qwen3 8B About 7 GB	~85 tok/s Runs great	~85 tok/s Runs great
Gemma 3 12B About 10 GB	~54 tok/s Runs great	~54 tok/s Runs great
Qwen3 14B About 11 GB	~47 tok/s Runs great	~47 tok/s Runs great
gpt-oss 20B About 16 GB	~140 tok/s Runs great	~140 tok/s Runs great
Gemma 3 27B About 19 GB	~26 tok/s Usable	~26 tok/s Usable
Qwen3 30B-A3B About 21 GB	~140 tok/s Runs great	~140 tok/s Runs great
Qwen3 32B About 22 GB	~22 tok/s Usable	~22 tok/s Usable
Llama 3.3 70B About 45 GB	~10 tok/s Usable	~10 tok/s Usable
gpt-oss 120B About 67 GB	~97 tok/s Runs great	~97 tok/s Runs great
Qwen3 235B-A22B About 144 GB	Needs 144 GB	~21 tok/s Usable
DeepSeek R1 671B About 406 GB	Needs 406 GB	Needs 406 GB

Which should you get?

It depends. The Apple M2 Ultra fits bigger models; the Apple M1 Ultra runs the models both fit faster. Price matters as much as either number: check used prices on the builds page, where every parts list has a measured or estimated speed.

How we estimate

Both columns use the same formula: memory bandwidth divided by the weights read per token, calibrated on our own rack, with Ollama's default 4-bit sizes. Real results are usually within about a third. Measured either one? Send us the number.

Compare

Other matchups

Every comparison