What LLMs Can the Apple M4 Pro 24GB Run?

The Apple M4 Pro 24GB has 24 GB of unified memory (about 16.0 GB usable by the GPU) and 273 GB/s of bandwidth. It runs 28 of 58 models fully at Q4_K_M with 8K context, up to Gemma 4 26B-A4B, and generates about 32 tokens/s on an 8B model.

Apple · Apple Silicon · Released 2024-10 · Updated October 14, 2026

28 of 58 models fit

Apple M4 Pro 24GB (16.0 GB GPU-usable) at Q4_K_M with 8K context. Largest that fits fully: Gemma 4 26B-A4B.

ModelNeedsStatusSpeedBest quant that fits
gpt-oss-20b12.9 GBFits~65 tok/sMXFP4 (official)
Qwen 2.5 14B10.7 GBFits~18 tok/sQ6_K
DeepSeek R1 Distill Qwen 14B10.7 GBFits~18 tok/sQ6_K
Phi-4 14B10.6 GBFits~18 tok/sQ6_K
Qwen 3 14B10.5 GBFits~18 tok/sQ6_K
Ministral 3 14B9.7 GBFits~20 tok/sQ6_K
Mistral Nemo 12B9.0 GBFits~22 tok/sQ8_0
Gemma 3 12B8.8 GBFits~22 tok/sQ8_0
Ministral 3 8B6.7 GBFits~31 tok/sQ8_0
Qwen 3 8B6.6 GBFits~32 tok/sQ8_0
Qwen 3.5 9B6.4 GBFits~31 tok/sQ8_0
Llama 3.1 8B6.3 GBFits~32 tok/sQ8_0
DeepSeek R1 Distill Llama 8B6.3 GBFits~32 tok/sQ8_0
Mistral 7B v0.35.7 GBFits~36 tok/sQ8_0
Gemma 4 E4B5.7 GBFits~35 tok/sQ8_0
Qwen 2.5 7B5.6 GBFits~36 tok/sQ8_0
DeepSeek R1 Distill Qwen 7B5.6 GBFits~36 tok/sQ8_0
Qwen 3 4B4.2 GBFits~57 tok/sFP16 / BF16
Phi-4 Mini4.1 GBFits~61 tok/sFP16 / BF16
Gemma 4 E2B4.0 GBFits~56 tok/sFP16 / BF16
Gemma 3 4B3.7 GBFits~62 tok/sFP16 / BF16
Qwen 3.5 4B3.7 GBFits~64 tok/sFP16 / BF16
Ministral 3 3B3.5 GBFits~69 tok/sFP16 / BF16
Llama 3.2 3B3.5 GBFits~72 tok/sFP16 / BF16
Llama 3.2 1B1.7 GBFits~194 tok/sFP16 / BF16
Gemma 3 1B1.6 GBFits~261 tok/sFP16 / BF16
Mistral Small 3.1 24B15.6 GBTight fit~12 tok/sIQ4_XS
Gemma 4 26B-A4B15.6 GBTight fit~57 tok/sIQ4_XS
Kimi K2.6566 GBToo small––
DeepSeek R1 671B380 GBToo small––
Llama 3.1 405B234 GBToo small––
Qwen 3.5 397B-A17B226 GBToo small––
DeepSeek V4 Flash161 GBToo small––
Qwen 3 235B-A22B135 GBToo small––
MiniMax-M2133 GBToo small––
Qwen 3.5 122B-A10B70.2 GBToo small––
Mistral Small 4 119B68.2 GBToo small––
gpt-oss-120b67.2 GBToo small––
Llama 4 Scout64.0 GBToo small––
GLM-4.5-Air62.1 GBToo small––
Qwen3-Next 80B-A3B46.3 GBToo small––
Qwen 2.5 72B44.5 GBToo small––
Llama 3.3 70B43.2 GBToo small––
DeepSeek R1 Distill Llama 70B43.2 GBToo small––
Mixtral 8x7B28.0 GBToo small––
Qwen 2.5 32B21.4 GBToo small–Q2_K
Qwen 2.5 Coder 32B21.4 GBToo small–Q2_K
DeepSeek R1 Distill Qwen 32B21.4 GBToo small–Q2_K
Qwen 3 32B21.4 GBToo small–Q2_K
Qwen 3.5 35B-A3B21.0 GBToo small–Q2_K
Qwen 3.6 35B-A3B21.0 GBToo small–Q2_K
Gemma 4 31B19.8 GBToo small–Q2_K
Qwen 3 30B-A3B18.8 GBToo small–Q2_K
Qwen3-Coder 30B-A3B18.8 GBToo small–Q2_K
Gemma 3 27B17.6 GBToo small–Q3_K_M
Qwen 3.5 27B16.9 GBToo small–IQ4_XS
Qwen 3.6 27B16.9 GBToo small–IQ4_XS
Qwen 3.8 27B16.9 GBToo small–IQ4_XS

Apple M4 Pro 24GB for local AI

The base Mac mini M4 Pro. Only about 16 GB is GPU-usable by default, so treat it like a 16 GB GPU: 14B models comfortably, 24B models at Q4 with short context.

Among the 14 devices we track with about 16 GB of usable memory, the Apple M4 Pro 24GB is the slowest at generating text, because generation speed follows memory bandwidth. Others in the same class: RTX 5080 16GB, RTX 5070 Ti 16GB, RTX 4080 Super 16GB, RTX 4080 16GB. They all hold the same models; the faster ones simply answer sooner.

Our picks for the Apple M4 Pro 24GB

The model we would run first for each job, chosen from our recommendations by what fits fully with an 8K context. "Best quant" is the highest quality that fits; "max context" is at Q4_K_M or the native format.

TaskModelBest quantMax contextSpeed
Best all-rounderQwen 3 14BQ6_K37K~18 tok/s
Codinggpt-oss-20bMXFP4 (official)55K~61 tok/s
Fastest repliesgpt-oss-20bMXFP4 (official)55K~61 tok/s
Long documents (64K+)Qwen 3.5 9BQ8_0256K~31 tok/s
Images in, text outMinistral 3 14BQ6_K42K~20 tok/s
Image generationQwen-ImageGGUF Q5_K_S––
SpecValue
Memory24 GB unified (16.0 GB GPU-usable by default)
Memory bandwidth273 GB/s
ArchitectureApple Silicon
Released2024-10
Llama 3.1 8B Q4_K_M speed~32 tok/s
Largest model that fitsGemma 4 26B-A4B

Image and video models on the Apple M4 Pro 24GB

Peak VRAM with model swapping (ComfyUI default on smaller cards).

ModelFP8FP8 fitGGUF Q4Q4 fit
Stable Diffusion 1.51.8 GBFits1.7 GBFits
SDXL 1.04.1 GBFits4.1 GBFits
Stable Diffusion 3.5 Medium6.4 GBFits6.4 GBFits
Stable Diffusion 3.5 Large8.6 GBFits6.4 GBFits
FLUX.1 [dev]12.1 GBFits7.3 GBFits
FLUX.1 [schnell]12.1 GBFits7.3 GBFits
Qwen-Image20.4 GBToo small12.3 GBFits
Wan 2.2 A14B (T2V / I2V)18.8 GBToo small13.2 GBFits
Wan 2.2 TI2V 5B8.0 GBFits6.5 GBFits
HunyuanVideo18.3 GBToo small13.1 GBFits
FLUX.1 Kontext [dev]12.3 GBFits7.5 GBFits
FLUX.2 [dev]31.0 GBToo small23.6 GBToo small
Z-Image Turbo6.8 GBFits4.9 GBFits
Qwen-Image-Edit 250920.5 GBToo small12.4 GBFits
LTX-219.2 GBToo small12.6 GBFits

Upgrade path

The next step up is the Apple M5 32GB with 32 GB.

Apple M4 Pro 24GB for LLMs: common questions

Can the Apple M4 Pro 24GB run Llama 3.3 70B?

No. It needs about 43.2 GB at Q4_K_M, far more than the 16.0 GB available.

Can the Apple M4 Pro 24GB run Qwen 3 32B?

No. It needs about 21.4 GB at Q4_K_M, far more than the 16.0 GB available.

Can the Apple M4 Pro 24GB run Qwen 3 14B?

Yes. Qwen 3 14B needs about 10.5 GB at Q4_K_M with 8K context, and the best quality that fits is Q6_K.

How fast is the Apple M4 Pro 24GB for LLMs?

About 32 tokens per second for an 8B model at Q4_K_M, based on its 273 GB/s memory bandwidth. Larger models scale down roughly in proportion to their size.

How much of the 24 GB can the GPU use?

About 16.0 GB by default. You can raise it with sudo sysctl iogpu.wired_limit_mb=<MB> (resets on reboot), leaving at least 8 GB for macOS.

Related