Macs
Which AI models run on the M3 Pro (18 GB)?
The M3 Pro (18 GB) has 18 GB of memory, about 12.0 GB of it usable for a model, and 150 GB/s of memory bandwidth. 45 of the 110 models in our catalog run on it at 4-bit or better.
Data as of 2 November 2026 · Vendor specs
Models that run on it
| Model | Best build | Memory | Speed |
|---|---|---|---|
| Nemotron Nano 9B v29B | Q6_K | 8.6 GB | ~11 tok/s our est. |
| Qwen3.5 9B Instruct9B | Q8_0 | 12.0 GB | ~15 tok/s ModelFit est. |
| Qwen2.5 7B Instruct7B | Q4_K_M | 6.6 GB | ~28 tok/s ModelFit est. |
| Gemma 3 4B Instruct4B | Q8_0 | 5.0 GB | ~29 tok/s ModelFit est. |
| Ministral 3 14B13.5B | Q4_K_M | 10.3 GB | ~14 tok/s ModelFit est. |
| Mellum2.1 12B-A2.5B Thinking12.1B MoE | Q4_K_M | 9.3 GB | ~49 tok/s our est. |
| Mistral Nemo 12B12B | Q4_K_M | 10.7 GB | ~18 tok/s ModelFit est. |
| Gemma 3 12B Instruct12B | Q4_K_M | 10.7 GB | ~16 tok/s ModelFit est. |
| Gemma 4 12B12B | Q4_K_M | 9.2 GB | ~16 tok/s ModelFit est. |
| GLM-4-9B-Chat9.4B | Q8_0 | 11.2 GB | ~14 tok/s ModelFit est. |
| Gemma 2 9B Instruct9B | Q4_K_M | 8.2 GB | ~18 tok/s ModelFit est. |
| Ornith 1.0 9B9B | Q4_K_M | 6.7 GB | ~25 tok/s ModelFit est. |
| MiMo-V2.6-Distill-Qwen-9B9B | Q4_K_M | 6.9 GB | ~24 tok/s ModelFit est. |
| Jebadiah 9B9B | Q8_0 | 11.0 GB | ~15 tok/s ModelFit est. |
| Granite 4.2 8B8.8B | Q4_K_M | 6.1 GB | ~22 tok/s ModelFit est. |
| Granite Switch 4.1 8B8.8B | Q4_K_M | 7.6 GB | ~18 tok/s ModelFit est. |
| MiniCPM-V 4.5 8B8.7B | Q4_K_M | 6.8 GB | ~24 tok/s ModelFit est. |
| Ministral 3 8B8.4B | Q4_K_M | 7.1 GB | ~21 tok/s ModelFit est. |
| LFM2.5 8B-A1B8.3B MoE | Q4_K_M | 6.6 GB | ~73 tok/s ModelFit est. |
| Llama 3.1 8B Instruct8B | Q8_0 | 9.2 GB | ~15 tok/s ModelFit est. |
| Qwen3 8B8B | Q8_0 | 9.3 GB | ~15 tok/s ModelFit est. |
| Granite 4.1 8B Instruct8B | Q4_K_M | 6.6 GB | ~21 tok/s ModelFit est. |
| Mistral 7B Instruct7B | Q8_0 | 8.4 GB | ~16 tok/s ModelFit est. |
| Qwen2.5 Coder 7B7B | Q4_K_M | 6.6 GB | ~28 tok/s ModelFit est. |
| DeepSeek-R1 Distill Qwen 7B7B | Q8_0 | 8.7 GB | ~17 tok/s ModelFit est. |
| Gemma 4 E4B4.5B | Q8_0 | 8.7 GB | ~25 tok/s ModelFit est. |
| Qwen3.5 4B Instruct4B | Q8_0 | 5.4 GB | ~28 tok/s ModelFit est. |
| NVIDIA Nemotron 3.5 Bolt E4B4B | Q4_K_M | 9.1 GB | ~32 tok/s ModelFit est. |
| Phi-3 Mini 3.8B3.8B | Q4_K_M | 4.3 GB | ~26 tok/s our est. |
| Phi-4 Mini 3.8B3.8B | Q8_0 | 4.9 GB | ~27 tok/s ModelFit est. |
| Granite 4.2 3B3.7B | Q4_K_M | 3.2 GB | ~51 tok/s ModelFit est. |
| Ministral 3 3B3.4B | Q4_K_M | 4.1 GB | ~37 tok/s ModelFit est. |
| Qwen2.5 3B Instruct3B | Q4_K_M | 3.6 GB | ~59 tok/s ModelFit est. |
| Llama 3.2 3B Instruct3B | Q4_K_M | 3.6 GB | ~49 tok/s ModelFit est. |
| Granite 4.1 3B Instruct3B | Q4_K_M | 3.0 GB | ~51 tok/s ModelFit est. |
| Gemma 4 E2B2.3B | Q8_0 | 5.7 GB | ~59 tok/s ModelFit est. |
| Gemma 2 2B Instruct2B | Q4_K_M | 2.8 GB | ~57 tok/s ModelFit est. |
| Qwen3.5 2B Instruct2B | Q4_K_M | 2.8 GB | ~59 tok/s ModelFit est. |
| LFM3 1.7B1.73B | Q4_K_M | 2.1 GB | ~100 tok/s ModelFit est. |
| Qwen2.5 1.5B Instruct1.5B | Q4_K_M | 2.5 GB | ~104 tok/s ModelFit est. |
| Llama 3.2 1B Instruct1B | Q4_K_M | 2.0 GB | ~119 tok/s ModelFit est. |
| Gemma 3 1B Instruct1B | Q8_0 | 2.0 GB | ~105 tok/s ModelFit est. |
| Qwen3.5 0.8B Instruct0.8B | Q4_K_M | 1.8 GB | ~129 tok/s ModelFit est. |
| Qwen2.5 0.5B Instruct0.5B | Q4_K_M | 1.8 GB | ~216 tok/s ModelFit est. |
| SmolLM2 360M0.36B | Q4_K_M | 1.5 GB | ~124 tok/s ModelFit est. |
Speeds come from ModelFit's estimates (CC BY 4.0) where it covers the model, otherwise from our bandwidth formula. Method
Too big for this machine
Qwen2.5 14B Instruct (12.3 GB), Phi-3 Medium 14B (12.3 GB), Qwen2.5 Coder 14B (12.3 GB), Qwen3 14B (12.3 GB), DeepSeek-R1 Distill Qwen 14B (12.3 GB), Phi-4 14B (12.8 GB), gpt-oss-20b (15.1 GB), LFM2 24B-A2B Instruct (15.3 GB), Mistral Small 3.1 (16.4 GB), Devstral Small 2 24B (16.6 GB), Gemma 4 26B-A4B (17.4 GB), Qwen3.5 27B Instruct (17.4 GB) and more.
Questions
What is the best AI model for the M3 Pro (18 GB)?
Qwen3.5 9B Instruct at Q8_0: the largest model on our shortlist of current models that runs at 4-bit or better at a usable speed (about 15 tok/s, ModelFit est.).
How much memory can a model use on the M3 Pro (18 GB)?
About 12.0 GB of the 18 GB.
How fast is the M3 Pro (18 GB) for local LLMs?
Decode speed is limited by memory bandwidth: 150 GB/s. A model that reads 10 GB per token runs at under 15 tokens per second.