llmrun

Apple · M2 Max · Desktop

Best AI Models for Mac Studio M2 Max (32 GB)

Runs 27–35B models well; 70B doesn't fit in 28.5 GB.

32 GB unified28.5 GB for models400 GB/s21 TFLOPS FP162023
Memory
32 GB64 GB96 GB
Same chip in
Mac StudioMacBook Pro 14"MacBook Pro 16"

What LLMs can the Mac Studio M2 Max run?

#ModelSpeedllmrun ScoreQuant · Memory
1
Qwen3.8 27BAlibabaHow to run →
Q4_K_M · 17.9 GB
16.1 tok/sQ4_K_M · 17.9 GB
2
Qwen3.6 27BAlibaba
Q4_K_M · 17.9 GB
16.1 tok/sQ4_K_M · 17.9 GB
3
Gemma 4 26B A4B ITGoogle
Q4_K_M · 16.5 GB
63.2 tok/sQ4_K_M · 16.5 GB
4
GPT OSS 20BOpenAI
Q4_K_M · 13.1 GB
71.2 tok/sQ4_K_M · 13.1 GB
5
NVIDIA Nemotron 3.5 Lightning 30B A3B BF16NVIDIA
Q4_K_M · 19.3 GB
64.3 tok/sQ4_K_M · 19.3 GB
See all 311 models on the Mac Studio M2 Max →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 Max (32 GB)

The Mac Studio M2 Max is the fastest 32 GB machine: it runs 27–35B models well. macOS keeps about 3.5 GB, leaving 28.5 GB for models. With 21 TFLOPS it reads long prompts faster than most 32 GB machines: a 4K-token prompt takes 53 s. M2 Max generation, released 2023; the newer M3 Max writes -16% faster with 1.0× the compute.

Speed in 32 GB#1400 GB/s
Prompt reading#221 TFLOPS
For models28.5 GBmacOS keeps 3.5 GB
ChipM2 Maxreleased 2023

Mac Studio M2 Max strengths and weaknesses

Strengths

  • Runs 35B MoE models: Qwen3.6 35B A3B at 54.5 tok/s in 22 GB
  • Room for long context: Qwen3.8 27B at Q4 uses 17.9 GB, leaving 10.6 GB free

Weaknesses

  • 70B models don't fit: Llama 3.3 70B Instruct needs 51.3 GB at Q4 vs 28.5 GB usable here
  • Slow prompt reading: A 32K-token prompt on Qwen3.8 27B waits 7.1 min for the first token
  • No CUDA: vLLM, AWQ, GPTQ and FP8 builds don't run; use GGUF or MLX

Memory for models

28.5 GB of 32 GB
9B6.6 GB14B11.9 GB27–31B17.9 GB35B MoE22 GB
9B6.6 GB14B11.9 GB35B MoE22 GB
For models 28.5 GBmacOS keeps 3.5 GBModel size at Q4

70B models need 51.3 GB at Q4, more than the 28.5 GB available. The 64 GB config fits Gemma 4 31B IT →

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 Max (64 GB)Apple · 60.5 GB for models · M2 Max
same speed▲ 32 GBno price
Same familySame chip, 3× the memoryMac Studio M2 Max (96 GB)Apple · 92.5 GB for models · M2 Max
same speed▲ 64 GBno price
Same familyNewer machine, 57% slowerMac Mini M4 (32 GB)Apple · 28.5 GB for models · M4
▼ 57% speedsame memoryabout $999
Newer machine, as fastMac Studio M4 Max (36 GB)Apple · 32.5 GB for models · M4 Max
same speed▲ 4 GBabout $2,499
Reads prompts 17× faster, 24 GB for modelsRTX 4090NVIDIA · 24 GB VRAM · GPU
▲ 17× prompts▲ 177% speed▼ 4.5 GBabout $1,599rent $0.34/hr
Reads prompts 22× faster, 32 GB for modelsRTX 5090NVIDIA · 32 GB VRAM · GPU
▲ 22× prompts▲ 291% speed▲ 3.5 GBabout $5,376rent $0.69/hr

Mac Studio M2 Max tokens per second by model size

Output speed depends on memory bandwidth (400 GB/s). Grey rows need more than the 28.5 GB available.

  1. 9BQwen3.5 9B · Q4_K_M · 6.6 GB
    44.0tok/s
  2. 14BQwen3 14B · Q4_K_M · 11.9 GB
    29.5tok/s
  3. 27–31BQwen3.8 27B · Q4_K_M · 17.9 GB
    16.1tok/s
  4. 35B MoEQwen3.6 35B A3B · Q4_K_M · 22.0 GB
    54.5tok/s
  5. 70BLlama 3.3 70B Instruct · does not fit
    needs 51.3 GB
  6. 120B MoEGPT OSS 120B · does not fit
    needs 70.8 GB

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

Max generations: writing vs compute

M1 Max → M5 Max
Writing speed (tok/s)Compute (TFLOPS FP16)
1×2×3×4×M1 Max2022M2 Max2023M3 Max2023M4 Max2025M5 Max2026M1 Max · 15.6 TFLOPS FP16M1 Max · 54.5 tok/s · 400 GB/sM2 Max · 21.5 TFLOPS FP16M2 Max · 54.5 tok/s · 400 GB/sM3 Max · 21.5 TFLOPS FP16M3 Max · 45.9 tok/s · 300 GB/sM4 Max · 25.8 TFLOPS FP16M4 Max · 55.2 tok/s · 409.6 GB/sM5 Max · 54.4 TFLOPS FP16M5 Max · 58.8 tok/s · 460 GB/sCompute 3.5×Writing +8%
1×2×3×4×M12022M22023M32023M42025M52026M1 Max · 15.6 TFLOPS FP16M1 Max · 54.5 tok/s · 400 GB/sM2 Max · 21.5 TFLOPS FP16M2 Max · 54.5 tok/s · 400 GB/sM3 Max · 21.5 TFLOPS FP16M3 Max · 45.9 tok/s · 300 GB/sM4 Max · 25.8 TFLOPS FP16M4 Max · 55.2 tok/s · 409.6 GB/sM5 Max · 54.4 TFLOPS FP16M5 Max · 58.8 tok/s · 460 GB/sCompute 3.5×Writing +8%

Writing speed follows memory bandwidth and barely moved: 54.5 to 58.8 tok/s since the M1 Max. Compute, which sets how fast prompts are read, grew 3.5×. Indexed to the M1 Max; speeds on Qwen3.6 35B A3B, same 32 GB.

All 1106 models on the Mac Studio M2 Max

Featured only

Best quant that fits in 28.5 GB, ordered by llmrun Score. Fit grades how well it runs here, S best. How ranking works

LLM models compatible with Mac Studio M2 Max (32 GB) — ranked by Score, recency and Fit
ModelVRAMScoreFit
Athene v2 Chat
72.7B·IQ2_S·11.8 t/s·33K ctx
28.4 GBD19, Barely runs
Qwen2.5 32B
32.8B·Q4_K_M·13.7 t/s·131K ctx
24.3 GBB46, Decent
Qwen2.5 Coder 14B
14.8B·Q4_K_M·29.3 t/s·33K ctx
12.4 GBA70, Runs well
Qwen2.5 Coder 32B
32.8B·Q4_K_M·13.7 t/s·33K ctx
24.3 GBB46, Decent
Llama 3.1 Tulu 3 70B
70.6B·IQ2_S·12.2 t/s·131K ctx
27.7 GBD24, Barely runs
Deepseek Coder 33B Base
33.3B·Q4_K_M·13.4 t/s·16K ctx
24.5 GBB45, Decent
Hermes 3 Llama 3.1 70B
70.6B·IQ2_XS·12.6 t/s·131K ctx
26.8 GBC30, Usable
DeepSeek R1 Distill Llama 70B
70.6B·IQ2_S·12.2 t/s·131K ctx
27.7 GBD24, Barely runs
Llama 3.1 Tulu 3 70B DPO
70.6B·IQ2_S·12.2 t/s·131K ctx
27.7 GBD24, Barely runs
Deepseek Coder 1.3B Base
1.3B·Q4_K_M·189.2 t/s·16K ctx
4.3 GBS99, Runs great
Llama 3.3 70B Instruct Abliterated
70.6B·IQ2_S·12.2 t/s·131K ctx
27.7 GBD24, Barely runs
Deepseek Coder 6.7B Base
6.7B·Q4_K_M·51.7 t/s·16K ctx
12.9 GBA80, Runs well
Falcon 40B
41.8B·Q3_K_L·11.9 t/s
27.7 GBD24, Barely runs
StableBeluga2
70B·IQ2_M·11.4 t/s·4K ctx
25.3 GBC38, Usable
Qwen2.5 72B Instruct
72.7B·IQ2_S·11.8 t/s·33K ctx
28.4 GBD19, Barely runs

Mac Studio M2 Max specs

Specs

Chip
M2 Max
CPU cores
12
GPU cores
30
Unified memory
32 GB
For models
28.5 GB (macOS keeps 3.5 GB)
Memory bandwidth
400 GB/s
FP16 (dense)
21.5 TFLOPS
Neural Engine
16 TOPS
Released
2023-06-13

Mac Studio M2 Max FAQ

How much memory is available for AI on the Mac Studio M2 Max (32 GB)?

About 28.5 GB of the 32 GB. macOS keeps roughly 3.5 GB for itself and other apps.

What is the largest model the Mac Studio M2 Max (32 GB) can run?

Qwen3.6 35B A3B (36.0B) at Q4_K_M uses 22.0 GB, 77% of the available memory, and runs at about 54.5 tok/s.

Can the Mac Studio M2 Max (32 GB) run Qwen3.8 27B?

Yes. Qwen3.8 27B fits at Q4_K_M, using 17.9 GB of the 28.5 GB available, and it is the best model that runs well on this machine.

Can the Mac Studio M2 Max (32 GB) run 70B models?

No. Llama 3.3 70B Instruct needs 51.3 GB at Q4, more than the 28.5 GB available.

How long does the Mac Studio M2 Max (32 GB) take to read a long prompt?

On Qwen3.8 27B, about 53 s for a 4K-token prompt and 7.1 min for a 32K-token document before the first word appears.

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

An RTX PRO 4500 Blackwell (32 GB) rents from $0.34/hr.

24 GB GPUsAll MacsDesktops & mini PCsMac Studio M1 MaxMacBook Pro 14" M1 MaxMacBook Pro 16" M1 MaxJetson AGX Orin 32GBMacBook Pro 16" M4 MaxBest coding LLMsAll hardware