aicanrun

AI Models for MacBook Pro M3 Pro — What runs on 18GB

46 great · 3 slow · 17 won't fit
Chip
Apple M3 Pro
Memory bandwidth
150 GB/s
Unified memory
18 GB
Usable for models
~15 GB
Cooling
Active
Year
2023

Specs checked against manufacturer and public documentation on . Results below are estimates, not measurements.

What runs on the MacBook Pro M3 Pro

All 66 models at the best tracked quant for this 18GB configuration. Select any row for the full report and step-by-step setup guide.

ModelParamsQuantNeedsSpeedVerdict
Bonsai 27B (1-bit)27BQ1_05GB~19.7–27.6 tokens/s✓ Runs great
Ternary Bonsai 27B27BQ2_08.6GB~10.4–14.6 tokens/s✓ Runs great
Qwen3 14B14.8BQ4_K_M11.3GB~8.3–11.7 tokens/s✓ Runs great
Phi-4 14B14.7BQ4_K_M11.4GB~8.2–11.5 tokens/s✓ Runs great
Ministral 3 14B14BQ4_K_M10.4GB~9.1–12.8 tokens/s✓ Runs great
Gemma 3 12B12.2BQ4_K_M9.3GB~10.3–14.4 tokens/s✓ Runs great
Gemma 4 12B12BQ4_09GB~10.7–15 tokens/s✓ Runs great
Grug 12B12BQ4_K_M9.7GB~9.7–13.6 tokens/s✓ Runs great
grug 9B (ProCreations)9.4BQ4_K_M7.7GB~12.7–17.8 tokens/s✓ Runs great
Fara 1.5 9B9.4BQ4_K_M7.7GB~12.7–17.8 tokens/s✓ Runs great
Qwen 3.5 9B9BQ4_K_M7.4GB~13.2–18.4 tokens/s✓ Runs great
Ornith 1.0 9B9BQ4_K_M7.3GB~13.4–18.8 tokens/s✓ Runs great
Qwythos 9B v29BQ4_K_M7.4GB~13.2–18.4 tokens/s✓ Runs great
Tini Cybersec 8B-A1B8.5BQ4_K_M6.9GB~122.6–171.6 tokens/s✓ Runs great
Qwen3 8B8.2BQ4_K_M6.6GB~15–21 tokens/s✓ Runs great
Llama 3.1 8B8BQ4_K_M6.5GB~15.3–21.4 tokens/s✓ Runs great
Ministral 8B8BQ4_K_M6.5GB~15.3–21.4 tokens/s✓ Runs great
Ternary Bonsai 8B8BQ2_03.7GB~34.1–47.7 tokens/s✓ Runs great
Ministral 3 8B8BQ4_K_M6.8GB~14.4–20.2 tokens/s✓ Runs great
LFM2.5 8B-A1B8BQ4_K_M6.8GB~115.4–161.5 tokens/s✓ Runs great
DeepSeek R1 Distill 7B7.6BQ4_K_M6.3GB~16–22.3 tokens/s✓ Runs great
Mistral 7B v0.37.2BQ4_K_M5.9GB~17–23.9 tokens/s✓ Runs great
AREX Turbo 4B4.5BQ4_K_M4.2GB~25.9–36.2 tokens/s✓ Runs great
Fara 1.5 4B4.5BQ4_K_M4.2GB~25.9–36.2 tokens/s✓ Runs great
Gemma 3 4B4.3BQ4_K_M3.7GB~30–42 tokens/s✓ Runs great
Qwen3 4B4BQ4_K_M3.7GB~30–42 tokens/s✓ Runs great
Qwen 3.5 4B4BQ4_K_M3.9GB~27.8–38.9 tokens/s✓ Runs great
Gemma 4 E4B4BQ4_K_M6.3GB~15–21 tokens/s✓ Runs great
Ternary Bonsai 4B4BQ2_02.2GB~68.2–95.5 tokens/s✓ Runs great
Nemotron 3 Nano 4B4BQ4_K_M4GB~26.8–37.5 tokens/s✓ Runs great
Agents-A1 4B4BQ4_K_M3.9GB~27.8–38.9 tokens/s✓ Runs great
Phi-4 Mini 3.8B3.8BQ4_K_M3.7GB~30–42 tokens/s✓ Runs great
Llama 3.2 3B3.2BQ4_K_M3.1GB~37.5–52.5 tokens/s✓ Runs great
SmolLM3 3B3.1BQ4_K_M3GB~39.5–55.3 tokens/s✓ Runs great
Ministral 3 3B3BQ4_K_M3.2GB~35.7–50 tokens/s✓ Runs great
G9v3 3B3BQ4_K_M3GB~39.5–55.3 tokens/s✓ Runs great
Qwen 3.5 2B2BQ4_K_M2.3GB~57.7–80.8 tokens/s✓ Runs great
Gemma 4 E2B2BQ4_K_M4.2GB~24.2–33.9 tokens/s✓ Runs great
DeepSeek R1 Distill 1.5B1.8BQ4_K_M2.1GB~68.2–95.5 tokens/s✓ Runs great
Qwen3 1.7B1.7BQ8_02.8GB~41.7–58.3 tokens/s✓ Runs great
SmolLM2 1.7B1.7BQ4_K_M2.1GB~68.2–95.5 tokens/s✓ Runs great
Ternary Bonsai 1.7B1.7BQ2_01.4GB~150–210 tokens/s✓ Runs great
Llama 3.2 1B1.2BQ4_K_M1.7GB~93.8–131.3 tokens/s✓ Runs great
Gemma 3 1B1BQ4_K_M1.7GB~93.8–131.3 tokens/s✓ Runs great
OvisOCR2 0.8B0.8BQ4_K_M1.5GB~125–175 tokens/s✓ Runs great
Qwen3 0.6B0.6BQ8_01.5GB~125–175 tokens/s✓ Runs great
Salience 1.5 Flash30BQ2_Ksmaller version14.5GB~62–86.8 tokens/s! Runs, barely
Fara 1.5 27B27BQ2_Ksmaller version14.9GB~6.5–9.1 tokens/s! Runs, barely
GPT-OSS 20B21BMXFP415GB~36.2–50.6 tokens/s! Runs, barely
Inkling952.4BQ8_0967.1GB—✕ Won't fit
Ornith 1.0 397B397BQ4_K_M285.8GB—✕ Won't fit
Hunyuan 3 (Hy3)298.8BQ4_K_M213GB—✕ Won't fit
Laguna S 2.1118BQ4_K_M80.7GB—✕ Won't fit
Llama 3.3 70B70BQ4_K_M50.3GB—✕ Won't fit
Qwen 3.5 35B-A3B35BQ4_K_M26.4GB—✕ Won't fit
Qwen 3.6 35B-A3B35BQ4_K_M24.1GB—✕ Won't fit
Ornith 1.0 35B-A3B35BQ4_K_M25.5GB—✕ Won't fit
KAT-Coder V2.5 Dev35BQ4_K_M25.7GB—✕ Won't fit
Laguna XS 2.133BQ4_K_M24.4GB—✕ Won't fit
Qwen3 32B32.8BQ4_K_M23.9GB—✕ Won't fit
Gemma 4 31B31BQ4_K_M22.2GB—✕ Won't fit
Qwen3 30B A3B30.5BQ4_K_M22.5GB—✕ Won't fit
Nemotron 3 Nano 30B-A3B30BQ4_K_M28.7GB—✕ Won't fit
Qwen 3.5 27B27BQ4_K_M20.2GB—✕ Won't fit
Qwen 3.6 27B27BQ4_K_M18.7GB—✕ Won't fit
Gemma 4 26B-A4B26BQ4_017.7GB—✕ Won't fit

~ = bandwidth-based estimate · open a row to see the recommended app and exact steps

Other MacBook Pro M3 Pro memory options

Check 36GB

FAQ

What is the biggest local AI model the MacBook Pro M3 Pro · 18GB can run?

Salience 1.5 Flash is the largest model in our 66-model comparison that fits at Q2_K. It needs about 14.5GB inside our 15GB working-memory budget, with an estimated 62–86.8 tokens/s decode range.

How much of the 18GB unified memory is available to a local LLM?

We use a conservative 15GB working budget, leaving room for macOS, the inference app, and normal background activity. Memory pressure, context length, and other open apps can change the real limit.

Can the MacBook Pro M3 Pro · 18GB run Llama 3.1 8B?

Yes. At Q4_K_M, our estimate uses 6.5GB and lands around 15.3–21.4 tokens/s.

Can the MacBook Pro M3 Pro · 18GB run Qwen 3.6 27B?

Not at the recommended Q4_K_M quant. It needs about 18.7GB in this 4K-context estimate.

Which app should I use for local AI on the MacBook Pro M3 Pro?

For an exact GGUF repository and quant from this site, start with LM Studio's graphical Discover, download, load, and chat flow. Jan is the open-source GGUF alternative; Ollama is strongest when the exact model already has a trustworthy catalog package or you want coding/API integrations; Msty is useful for mixed GGUF, MLX, and document workflows. For the curated Bonsai 27B build, try Locally AI and first confirm that its catalog shows the model on this Mac.

Can I upgrade the MacBook Pro M3 Pro to more unified memory later?

No. Apple-silicon unified memory is integrated into the chip package and must be chosen at purchase. Storage upgrades or external SSDs do not increase the memory an LLM can use.

Are these MacBook Pro M3 Pro local AI speeds measured?

No. Every speed on this page is a formula range based on Apple’s published memory bandwidth, model size, active parameters, and a cooling-aware efficiency range. App, backend, context length, and thermals can change real results.