llmrun

Apple · M2 · Mini PC

Best AI Models for Mac Mini M2 (24 GB)

Runs 27–31B models at 4 tok/s; 9B and 14B fit but are slow (11 and 7 tok/s), and 70B doesn't fit in 20.5 GB.

24 GB unified20.5 GB for models100 GB/s7 TFLOPS FP162023
Memory
8 GB16 GB24 GB
Same chip in
Mac MiniMacBook Air 13"MacBook Air 15"MacBook Pro 13"

Top models by task

#ModelSpeedllmrun ScoreQuant · Memory
1
Gemma 4 26B A4B ITGoogleHow to run →
Q4_K_M · 16.5 GB
22.1 tok/sQ4_K_M · 16.5 GB
2
Qwen3.5 9BAlibaba
Q4_K_M · 6.6 GB
11.0 tok/sQ4_K_M · 6.6 GB
3
GPT OSS 20BOpenAI
Q4_K_M · 13.1 GB
23.9 tok/sQ4_K_M · 13.1 GB
4
Gemma 3 4B ITGoogle
Q4_K_M · 5.0 GB
24.6 tok/sQ4_K_M · 5.0 GB
5
Qwen3.5 4BAlibaba
Q4_K_M · 3.4 GB
21.5 tok/sQ4_K_M · 3.4 GB
See all 293 models on the Mac Mini M2 →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 Mini M2 (24 GB)

The Mac Mini M2 is a 24 GB mini PC with more memory than speed: it runs 27–31B models at 4 tok/s. macOS keeps about 3.5 GB, leaving 20.5 GB for models. Prompt reading is its weak spot: a 4K-token prompt waits 25 s before the first word on Gemma 4 26B A4B IT. M2 generation, released 2023; the newer M3 writes 2% faster with 1.0× the compute.

Speed in 24 GB#6100 GB/s
Prompt reading#57 TFLOPS
For models20.5 GBmacOS keeps 3.5 GB
ChipM2released 2023

Mac Mini M2 strengths and weaknesses

Strengths

  • Runs 35B MoE models: Qwen3.6 35B A3B at 22.1 tok/s in 19.8 GB

Weaknesses

  • 70B models don't fit: Llama 3.3 70B Instruct needs 51.3 GB at Q4 vs 20.5 GB usable here
  • Slow on large dense models: Qwen3.8 27B at 4.0 tok/s
  • Slow prompt reading: A 32K-token prompt on Gemma 4 26B A4B IT waits 3.4 min for the first token
  • No CUDA: vLLM, AWQ, GPTQ and FP8 builds don't run; use GGUF or MLX

Memory for models

20.5 GB of 24 GB
9B6.6 GB14B11.9 GB27–31B17.9 GB35B MoE19.8 GB
9B6.6 GB14B11.9 GB35B MoE19.8 GB
For models 20.5 GBmacOS keeps 3.5 GBModel size at Q4

70B models need 51.3 GB at Q4, more than the 20.5 GB available.

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, 8 GB less memoryMac Mini M2 (16 GB)Apple · 12.5 GB for models · M2
▲ 51% speed▼ 8 GBno price
Same family110% faster, same memoryMac Mini M4 Pro (24 GB)Apple · 20.5 GB for models · M4 Pro
▲ 110% speedsame memoryabout $1,599
16% faster, same memoryMacBook Air 13" M4 (24 GB)Apple · 20.5 GB for models · M4
▲ 16% speedsame memoryabout $1,399
29% faster, 3.5 GB more memoryMacBook Pro 14 M5Apple · 24 GB for models · M5
▲ 29% speed▲ 3.5 GBabout $1,839
Reads prompts 4× faster, 32.5 GB for modelsMac Studio M4 Max (36 GB)Apple · 32.5 GB for models · M4 Max
▲ 4× prompts▲ 150% speed▲ 12 GBabout $2,499
Reads prompts 52× faster, 24 GB for modelsRTX 4090NVIDIA · 24 GB VRAM · GPU
▲ 52× prompts▲ 583% speed▲ 3.5 GBabout $1,599rent $0.34/hr

Mac Mini M2 tokens per second by model size

Output speed depends on memory bandwidth (100 GB/s). Grey rows need more than the 20.5 GB available.

  1. 9BQwen3.5 9B · Q4_K_M · 6.6 GB
    11.0tok/s
  2. 14BQwen3 14B · Q4_K_M · 11.9 GB slow
    7.4tok/s
  3. 27–31BQwen3.8 27B · Q4_K_M · 17.9 GB slow
    4.0tok/s
  4. 35B MoEQwen3.6 35B A3B · IQ4_XS · 19.8 GB tight · IQ4_XS
    22.1tok/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

Chip generations: writing vs compute

M1 → M5
Writing speed (tok/s)Compute (TFLOPS FP16)
1×2×3×4×M12020M22023M32024M42024M52026M1 · 5.2 TFLOPS FP16M1 · 24.9 tok/s · 68.3 GB/sM2 · 7.2 TFLOPS FP16M2 · 22.1 tok/s · 100 GB/sM3 · 7 TFLOPS FP16M3 · 22.6 tok/s · 102.4 GB/sM4 · 8.6 TFLOPS FP16M4 · 25.7 tok/s · 120 GB/sM5 · 16 TFLOPS FP16M5 · 31.1 tok/s · 153.6 GB/sCompute 3.1×Writing +25%
1×2×3×4×M12020M22023M32024M42024M52026M1 · 5.2 TFLOPS FP16M1 · 24.9 tok/s · 68.3 GB/sM2 · 7.2 TFLOPS FP16M2 · 22.1 tok/s · 100 GB/sM3 · 7 TFLOPS FP16M3 · 22.6 tok/s · 102.4 GB/sM4 · 8.6 TFLOPS FP16M4 · 25.7 tok/s · 120 GB/sM5 · 16 TFLOPS FP16M5 · 31.1 tok/s · 153.6 GB/sCompute 3.1×Writing +25%

Writing speed follows memory bandwidth and barely moved: 24.9 to 31.1 tok/s since the M1. Compute, which sets how fast prompts are read, grew 3.1×. Indexed to the M1; speeds on Qwen3.6 35B A3B, same 16 GB.

All 1068 models on the Mac Mini M2

Featured only

Best quant that fits in 20.5 GB, ordered by llmrun Score. Fit grades how well it runs here, S best. How ranking works

LLM models compatible with Mac Mini M2 (24 GB) — ranked by Score, recency and Fit
ModelVRAMScoreFit
Kimi Linear 48B A3B Base
49.1B·IQ2_S·45.9 t/s
19.7 GBB53, Decent
Granite Guardian 3.3 8B
8.2B·Q4_K_M·12.6 t/s·131K ctx
7.9 GBB51, Decent
XiYanSQL QwenCoder 32B 2504
32B·Q3_K_M·4.3 t/s·33K ctx
20.2 GBF11, Won't fit
Internlm Chat 7B
7B·FP16·4.5 t/s·2K ctx
15.4 GBC41, Usable
WhiteRabbitNeo 13B V1
13B·Q3_K_L·8.1 t/s·16K ctx
20.4 GBD16, Barely runs
Pollux Judge 32B
32.8B·Q2_K·4.7 t/s·33K ctx
18.5 GBD27, Barely runs
Nous Hermes 2 Mixtral 8x7B DPO
46.7B·IQ2_M·14.1 t/s·33K ctx
18.2 GBC41, Usable
Llama Guard 3 8B
8.0B·BF16·4.0 t/s
20.2 GBF10, Won't fit
Granite Guardian 3.2 8B Factuality Detection
8.2B·Q4_K_M·12.6 t/s·131K ctx
7.9 GBB51, Decent
Yi 34B
34.4B·IQ4_XS·3.6 t/s·4K ctx
19.8 GBF13, Won't fit
Nemotron H 8B Reasoning 128K
8.1B·BF16·3.9 t/s
20.4 GBF9, Won't fit
Qwen3 42B A3B 2507 Thinking Abliterated Uncensored TOTAL RECALL v2 Medium MASTER CODER
42.4B·Q2_K·24.4 t/s·262K ctx
19.4 GBC42, Usable
Yi 34B Chat
34.4B·IQ4_XS·3.6 t/s·4K ctx
19.8 GBF13, Won't fit
Qwen2.5 Coder 32B
32.8B·IQ3_M·4.5 t/s·33K ctx
19.3 GBD20, Barely runs
Qwen1.5 14B
14.2B·IQ3_M·8.4 t/s·33K ctx
20.1 GBD18, Barely runs

Mac Mini M2 specs

Specs

Chip
M2
CPU cores
8
GPU cores
10
Unified memory
24 GB
For models
20.5 GB (macOS keeps 3.5 GB)
Memory bandwidth
100 GB/s
FP16 (dense)
7.2 TFLOPS
Neural Engine
16 TOPS
Released
2023-01-24

Mac Mini M2 FAQ

How much memory is available for AI on the Mac Mini M2 (24 GB)?

About 20.5 GB of the 24 GB. macOS keeps roughly 3.5 GB for itself and other apps.

What is the largest model the Mac Mini M2 (24 GB) can run?

Qwen3.6 35B A3B (36.0B) at IQ4_XS uses 19.8 GB, 97% of the available memory, and runs at about 22.1 tok/s.

Can the Mac Mini M2 (24 GB) run 70B models?

No. Llama 3.3 70B Instruct needs 51.3 GB at Q4, more than the 20.5 GB available.

How long does the Mac Mini M2 (24 GB) take to read a long prompt?

On Gemma 4 26B A4B IT, about 25 s for a 4K-token prompt and 3.4 min for a 32K-token document before the first word appears.

Is renting a GPU cheaper than buying the Mac Mini M2?

An RTX 3090 (24 GB) rents from $0.16/hr.

16 GB GPUsAll MacsDesktops & mini PCsMac Mini M5 ProMacBook Pro 14" M5 ProMacBook Pro 16" M5 ProMacBook Pro 14" M4 ProMacBook Pro 16" M4 ProMac Mini M4Best coding LLMsAll hardware