Apple · M2 Ultra · Desktop

Best AI Models for Mac Studio M2 Ultra (64 GB)

Runs 27–35B models well and fits 120B MoE; dense 70B fits but is slow at 12 tok/s.

64 GB unified60.5 GB for models800 GB/s43 TFLOPS FP162023

Memory

Same chip in

What LLMs can the Mac Studio M2 Ultra run?

#	Model	Speed	Quant · Memory
1	Qwen3.8 27BAlibabaHow to run → Q4_K_M · 17.9 GB	32.1 tok/s	Q4_K_M · 17.9 GB
2	Qwen3.6 27BAlibaba Q4_K_M · 17.9 GB	32.1 tok/s	Q4_K_M · 17.9 GB
3	Gemma 4 31B ITGoogle Q4_K_M · 21.7 GB	27.5 tok/s	Q4_K_M · 21.7 GB
4	Gemma 4 26B A4B ITGoogle Q4_K_M · 16.5 GB	91.7 tok/s	Q4_K_M · 16.5 GB
5	NVIDIA Nemotron 3.5 Lightning 30B A3B BF16NVIDIA Q4_K_M · 19.3 GB	77.0 tok/s	Q4_K_M · 19.3 GB

#	Model	Speed	Quant · Memory
1	Qwen3.8 27BAlibabaHow to run → Q4_K_M · 17.9 GB	32.1 tok/s	Q4_K_M · 17.9 GB
2	Gemma 4 31B ITGoogle Q4_K_M · 21.7 GB	27.5 tok/s	Q4_K_M · 21.7 GB
3	Qwen3.6 35B A3BAlibaba Q4_K_M · 22.0 GB	75.9 tok/s	Q4_K_M · 22.0 GB
4	Gemma 4 26B A4B ITGoogle Q4_K_M · 16.5 GB	91.7 tok/s	Q4_K_M · 16.5 GB
5	NVIDIA Nemotron 3.5 Lightning 30B A3B BF16NVIDIA Q4_K_M · 19.3 GB	77.0 tok/s	Q4_K_M · 19.3 GB

#	Model	Speed	Quant · Memory
1	Qwen3.8 27BAlibabaHow to run → Q4_K_M · 17.9 GB	32.1 tok/s	Q4_K_M · 17.9 GB
2	Qwen3.5 27BAlibaba Q4_K_M · 17.9 GB	32.1 tok/s	Q4_K_M · 17.9 GB
3	Gemma 4 31B ITGoogle Q4_K_M · 21.7 GB	27.5 tok/s	Q4_K_M · 21.7 GB
4	Gemma 4 26B A4B ITGoogle Q4_K_M · 16.5 GB	91.7 tok/s	Q4_K_M · 16.5 GB
5	NVIDIA Nemotron 3.5 Lightning 30B A3B BF16NVIDIA Q4_K_M · 19.3 GB	77.0 tok/s	Q4_K_M · 19.3 GB

#	Model	Speed	Math Score	Quant · Memory
1	Qwen3.8 27BAlibabaHow to run → Q4_K_M · 17.9 GB	32.1 tok/s		Q4_K_M · 17.9 GB
2	Qwen3.6 27BAlibaba Q4_K_M · 17.9 GB	32.1 tok/s		Q4_K_M · 17.9 GB
3	Qwen3.6 35B A3BAlibaba Q4_K_M · 22.0 GB	75.9 tok/s		Q4_K_M · 22.0 GB
4	Gemma 4 31B ITGoogle Q4_K_M · 21.7 GB	27.5 tok/s	overall	Q4_K_M · 21.7 GB
5	NVIDIA Nemotron 3.5 Lightning 30B A3B BF16NVIDIA Q4_K_M · 19.3 GB	77.0 tok/s		Q4_K_M · 19.3 GB

#	Model	Speed ↓	Quant · Memory
1	GPT OSS 20BOpenAIHow to run → Q4_K_M · 13.1 GB	106 tok/s	Q4_K_M · 13.1 GB
2	Gemma 4 26B A4B ITGoogle Q4_K_M · 16.5 GB	91.7 tok/s	Q4_K_M · 16.5 GB
3	Qwen3.5 9BAlibaba Q4_K_M · 6.6 GB	88.1 tok/s	Q4_K_M · 6.6 GB
4	NVIDIA Nemotron 3.5 Lightning 30B A3B BF16NVIDIA Q4_K_M · 19.3 GB	77.0 tok/s	Q4_K_M · 19.3 GB
5	Qwen3.6 35B A3BAlibaba Q4_K_M · 22.0 GB	75.9 tok/s	Q4_K_M · 22.0 GB

See all 338 models on the Mac Studio M2 Ultra →Bars = traction for models without a Score yet
Task rankings use that task's benchmarks, topped up from Overall when short; bars mark well-known models with no Score yet.

About the Mac Studio M2 Ultra (64 GB)

The Mac Studio M2 Ultra is the fastest 64 GB machine: it runs 27–35B models well and fits 120B MoE. macOS keeps about 3.5 GB, leaving 60.5 GB for models. With 43 TFLOPS it reads long prompts faster than most 64 GB machines: a 4K-token prompt takes 26 s. M2 Ultra generation, released 2023; the newer M3 Ultra writes 1% faster with 1.0× the compute.

Speed in 64 GB#1800 GB/s

Prompt reading#243 TFLOPS

For models60.5 GBmacOS keeps 3.5 GB

ChipM2 Ultrareleased 2023

Mac Studio M2 Ultra strengths and weaknesses

Strengths

Runs 120B MoE models: GPT OSS 120B at 80.8 tok/s in 59.1 GB
Room for long context: Qwen3.8 27B at Q4 uses 17.9 GB, leaving 42.6 GB free

Weaknesses

Slow prompt reading: A 32K-token prompt on Qwen3.8 27B waits 3.5 min for the first token
No CUDA: vLLM, AWQ, GPTQ and FP8 builds don't run; use GGUF or MLX

Memory for models

60.5 GB of 64 GB

9B6.6 GB14B11.9 GB35B MoE22 GB70B51.3 GB120B MoE59.1 GB

35B MoE22 GB120B MoE59.1 GB

For models 60.5 GBmacOS keeps 3.5 GBModel size at Q4

Everything on the size ladder fits. Two 27B-class models together use 35.7 GB and leave 24.8 GB free.

Similar devices and alternatives

Other memory sizes of the same machine first, then machines and GPUs people weigh against it, each with the one reason it is worth a look. Speeds on Qwen3.6 35B A3B.

Same chip, 2× the memoryMac Studio M2 Ultra (128 GB)Apple · 124.5 GB for models · M2 Ultra

same speed▲ 64 GBno price

Same family Same chip, 3× the memoryMac Studio M2 Ultra (192 GB)Apple · 188.5 GB for models · M2 Ultra

same speed▲ 128 GB$5,599

Same family Newer machine, 15% slowerMac Studio M4 Max (64 GB)Apple · 60.5 GB for models · M4 Max

▼ 15% speedsame memory$2,699

Newer machine, as fastMac Studio M3 Ultra (96 GB)Apple · 72 GB for models · M3 Ultra

same speed▲ 11.5 GB$3,999

2.0× faster if your models fit in 24 GBRTX 4090NVIDIA · 24 GB VRAM · GPU

▲ 99% speed▼ 36.5 GB$1,599rent $0.34/hr

2.8× faster if your models fit in 32 GBRTX 5090NVIDIA · 32 GB VRAM · GPU

▲ 181% speed▼ 28.5 GB$5,376rent $0.59/hr

Mac Studio M2 Ultra tokens per second by model size

Output speed depends on memory bandwidth (800 GB/s). Grey rows need more than the 60.5 GB available.

9BQwen3.5 9B · Q4_K_M · 6.6 GB
88.1tok/s
14BQwen3 14B · Q4_K_M · 11.9 GB
58.9tok/s
27–31BQwen3.8 27B · Q4_K_M · 17.9 GB
32.1tok/s
35B MoEQwen3.6 35B A3B · Q4_K_M · 22.0 GB
75.9tok/s
70BLlama 3.3 70B Instruct · Q4_K_M · 51.3 GB
12.0tok/s
120B MoEGPT OSS 120B · Q4_0 · 59.1 GB tight · Q4_0
80.8tok/s

Estimated speeds: Q4_K_M, 4K context, one user. How we estimate

Ultra generations: writing vs compute

M1 Ultra → M3 Ultra

Writing speed (tok/s)Compute (TFLOPS FP16)

Writing speed follows memory bandwidth and barely moved: 75.9 to 76.6 tok/s since the M1 Ultra. Compute, which sets how fast prompts are read, grew 1.4×. Indexed to the M1 Ultra; speeds on Qwen3.6 35B A3B, same 64 GB.

All 1168 models on the Mac Studio M2 Ultra

Featured only

Best quant that fits in 60.5 GB, ordered by llmrun Score. Fit grades how well it runs here, S best. How ranking works

LLM models compatible with Mac Studio M2 Ultra (64 GB) — ranked by Score, recency and Fit
Model	Quant	VRAM	Speed	Context	Fit
Qwen2.5 Coder 1.5B1.5BChatCodeFunctions 1.5B·Q4_K_M·437.5 t/s·33K ctx	Q4_K_M	1.7 GB3%	437.5 t/s	33K	S100, Runs great
Meta Llama 3 70B Instruct70.6BChat 70.6B·Q4_K_M·12.9 t/s·8K ctx	Q4_K_M	45.3 GB75%	12.9 t/s	8K	B52, Decent
Gemma 7B IT8.5BChat 8.5B·BF16·29.8 t/s	BF16	21.4 GB35%	29.8 t/s	—	A71, Runs well
Devstral 2 123B Instruct 2512125.0BChatFunctions 125.0B·Q2_K·10.3 t/s·262K ctx	Q2_K	59.3 GB98%	10.3 t/s	262K	D21, Barely runs
Internlm 20B20BChat 20B·BF16·13.1 t/s·4K ctx	BF16	45.3 GB75%	13.1 t/s	4K	B52, Decent
Qwen1.5 32B32.5BChat 32.5B·Q4_K_M·27.5 t/s·33K ctx	Q4_K_M	24.1 GB40%	27.5 t/s	33K	A68, Runs well
Granite Guardian 3.2 8B Factuality Detection8.2BChat 8.2B·Q4_K_M·101.1 t/s·131K ctx	Q4_K_M	7.9 GB13%	101.1 t/s	131K	S92, Runs great
Qwen2.5 Coder 0.5B494MChatCode 494M·Q4_K_M·903.2 t/s·33K ctx	Q4_K_M	0.8 GB1%	903.2 t/s	33K	S100, Runs great
Qwen2.5 32B32.8BChat 32.8B·Q4_K_M·27.3 t/s·131K ctx	Q4_K_M	24.3 GB40%	27.3 t/s	131K	A68, Runs well
Llama 2 13B HF13.0BChat 13.0B·Q4_K_M·65.2 t/s	Q4_K_M	11.5 GB19%	65.2 t/s	—	A84, Runs well
XORTRON.CriminalComputing.LARGE.2026.3122.6BChat 122.6B·Q2_K·10.5 t/s·131K ctx	Q2_K	58.3 GB96%	10.5 t/s	131K	D24, Barely runs
Qwen1.5 1.8B1.8BChat 1.8B·Q4_K_M·311.1 t/s·33K ctx	Q4_K_M	4.6 GB8%	311.1 t/s	33K	S100, Runs great
Qwen2.5 Coder 32B32.8BChatCodeFunctions 32.8B·Q4_K_M·27.3 t/s·33K ctx	Q4_K_M	24.3 GB40%	27.3 t/s	33K	A68, Runs well
Deepseek Coder 33B Base33.3BChatCode 33.3B·Q4_K_M·26.9 t/s·16K ctx	Q4_K_M	24.5 GB40%	26.9 t/s	16K	A68, Runs well
Internlm Chat 7B7BChat 7B·FP16·36.4 t/s·2K ctx	FP16	15.4 GB25%	36.4 t/s	2K	A73, Runs well

Mac Studio M2 Ultra specs

Specs

Chip: M2 Ultra
CPU cores: 24
GPU cores: 60
Unified memory: 64 GB
For models: 60.5 GB (macOS keeps 3.5 GB)
Memory bandwidth: 800 GB/s
FP16 (dense): 43.0 TFLOPS
Neural Engine: 32 TOPS
Released: 2023-06-13

Mac Studio M2 Ultra FAQ

How much memory is available for AI on the Mac Studio M2 Ultra (64 GB)?

About 60.5 GB of the 64 GB. macOS keeps roughly 3.5 GB for itself and other apps.

What is the largest model the Mac Studio M2 Ultra (64 GB) can run?

GPT OSS 120B (116.8B) at Q4_0 uses 59.1 GB, 98% of the available memory, and runs at about 80.8 tok/s.

Can the Mac Studio M2 Ultra (64 GB) run Qwen3.8 27B?

Yes. Qwen3.8 27B fits at Q4_K_M, using 17.9 GB of the 60.5 GB available, and it is the best model that runs well on this machine.

How long does the Mac Studio M2 Ultra (64 GB) take to read a long prompt?

On Qwen3.8 27B, about 26 s for a 4K-token prompt and 3.5 min for a 32K-token document before the first word appears.

Is renting a GPU cheaper than buying the Mac Studio M2 Ultra?

An RTX PRO 6000 Blackwell Max-Q Workstation Edition (96 GB) rents from $1.22/hr.

48 GB GPUs All Macs Desktops & mini PCs Mac Studio M1 Ultra Mac Studio M5 Max MacBook Pro 14" M5 Max MacBook Pro 16" M5 Max Best coding LLMs All hardware