Macs
Which AI models run on the M2 (24 GB)?
The M2 (24 GB) has 24 GB of memory, about 16.0 GB of it usable for a model, and 100 GB/s of memory bandwidth. 53 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 |
|---|---|---|---|
| gpt-oss-20b21B MoE | MXFP4 | 15.1 GB | ~20 tok/s ModelFit est. |
| Nemotron Nano 9B v29B | Q4_K_M | 6.6 GB | ~10 tok/s our est. |
| Qwen3.5 9B Instruct9B | Q4_K_M | 8.2 GB | ~16 tok/s ModelFit est. |
| Qwen2.5 7B Instruct7B | Q4_K_M | 6.6 GB | ~18 tok/s ModelFit est. |
| Gemma 3 4B Instruct4B | Q8_0 | 5.0 GB | ~19 tok/s ModelFit est. |
| LFM2 24B-A2B Instruct24B MoE | Q4_K_M | 15.3 GB | ~35 tok/s ModelFit est. |
| Qwen2.5 14B Instruct14B | Q4_K_M | 12.3 GB | ~8.8 tok/s slow ModelFit est. |
| Phi-3 Medium 14B14B | Q4_K_M | 12.3 GB | ~5.0 tok/s slow our est. |
| Phi-4 14B14B | Q4_K_M | 12.8 GB | ~8.5 tok/s slow ModelFit est. |
| Qwen2.5 Coder 14B14B | Q4_K_M | 12.3 GB | ~8.8 tok/s slow ModelFit est. |
| Qwen3 14B14B | Q4_K_M | 12.3 GB | ~8.7 tok/s slow ModelFit est. |
| DeepSeek-R1 Distill Qwen 14B14B | Q4_K_M | 12.3 GB | ~8.8 tok/s slow ModelFit est. |
| Ministral 3 14B13.5B | Q4_K_M | 10.3 GB | ~8.8 tok/s slow ModelFit est. |
| Mellum2.1 12B-A2.5B Thinking12.1B MoE | Q4_K_M | 9.3 GB | ~33 tok/s our est. |
| Mistral Nemo 12B12B | Q4_K_M | 10.7 GB | ~11 tok/s ModelFit est. |
| Gemma 3 12B Instruct12B | Q4_K_M | 10.7 GB | ~10 tok/s ModelFit est. |
| Gemma 4 12B12B | Q4_K_M | 9.2 GB | ~10 tok/s ModelFit est. |
| GLM-4-9B-Chat9.4B | Q4_K_M | 7.5 GB | ~14 tok/s ModelFit est. |
| Gemma 2 9B Instruct9B | Q4_K_M | 8.2 GB | ~11 tok/s ModelFit est. |
| Ornith 1.0 9B9B | Q4_K_M | 6.7 GB | ~16 tok/s ModelFit est. |
| MiMo-V2.6-Distill-Qwen-9B9B | Q4_K_M | 6.9 GB | ~15 tok/s ModelFit est. |
| Jebadiah 9B9B | Q5_K_M | 7.8 GB | ~14 tok/s ModelFit est. |
| Granite 4.2 8B8.8B | Q4_K_M | 6.1 GB | ~14 tok/s ModelFit est. |
| Granite Switch 4.1 8B8.8B | Q4_K_M | 7.6 GB | ~11 tok/s ModelFit est. |
| MiniCPM-V 4.5 8B8.7B | Q4_K_M | 6.8 GB | ~15 tok/s ModelFit est. |
| Ministral 3 8B8.4B | Q4_K_M | 7.1 GB | ~13 tok/s ModelFit est. |
| LFM2.5 8B-A1B8.3B MoE | Q4_K_M | 6.6 GB | ~46 tok/s ModelFit est. |
| Llama 3.1 8B Instruct8B | Q5_K_M | 9.2 GB | ~14 tok/s ModelFit est. |
| Qwen3 8B8B | Q4_K_M | 7.7 GB | ~15 tok/s ModelFit est. |
| Granite 4.1 8B Instruct8B | Q4_K_M | 6.6 GB | ~14 tok/s ModelFit est. |
| Mistral 7B Instruct7B | Q8_0 | 8.4 GB | ~10 tok/s ModelFit est. |
| Qwen2.5 Coder 7B7B | Q4_K_M | 6.6 GB | ~18 tok/s ModelFit est. |
| DeepSeek-R1 Distill Qwen 7B7B | Q8_0 | 8.7 GB | ~11 tok/s ModelFit est. |
| Gemma 4 E4B4.5B | Q8_0 | 8.7 GB | ~16 tok/s ModelFit est. |
| Qwen3.5 4B Instruct4B | Q8_0 | 5.4 GB | ~18 tok/s ModelFit est. |
| NVIDIA Nemotron 3.5 Bolt E4B4B | Q4_K_M | 9.1 GB | ~21 tok/s ModelFit est. |
| Phi-3 Mini 3.8B3.8B | Q4_K_M | 4.3 GB | ~17 tok/s our est. |
| Phi-4 Mini 3.8B3.8B | Q8_0 | 4.9 GB | ~17 tok/s ModelFit est. |
| Granite 4.2 3B3.7B | Q4_K_M | 3.2 GB | ~32 tok/s ModelFit est. |
| Ministral 3 3B3.4B | Q4_K_M | 4.1 GB | ~24 tok/s ModelFit est. |
| Qwen2.5 3B Instruct3B | Q4_K_M | 3.6 GB | ~38 tok/s ModelFit est. |
| Llama 3.2 3B Instruct3B | Q4_K_M | 3.6 GB | ~32 tok/s ModelFit est. |
| Granite 4.1 3B Instruct3B | Q4_K_M | 3.0 GB | ~32 tok/s ModelFit est. |
| Gemma 4 E2B2.3B | Q8_0 | 5.7 GB | ~37 tok/s ModelFit est. |
| Gemma 2 2B Instruct2B | Q4_K_M | 2.8 GB | ~36 tok/s ModelFit est. |
| Qwen3.5 2B Instruct2B | Q4_K_M | 2.8 GB | ~38 tok/s ModelFit est. |
| LFM3 1.7B1.73B | Q4_K_M | 2.1 GB | ~65 tok/s ModelFit est. |
| Qwen2.5 1.5B Instruct1.5B | Q4_K_M | 2.5 GB | ~67 tok/s ModelFit est. |
| Llama 3.2 1B Instruct1B | Q4_K_M | 2.0 GB | ~77 tok/s ModelFit est. |
| Gemma 3 1B Instruct1B | Q8_0 | 2.0 GB | ~68 tok/s ModelFit est. |
| Qwen3.5 0.8B Instruct0.8B | Q4_K_M | 1.8 GB | ~83 tok/s ModelFit est. |
| Qwen2.5 0.5B Instruct0.5B | Q4_K_M | 1.8 GB | ~143 tok/s ModelFit est. |
| SmolLM2 360M0.36B | Q4_K_M | 1.5 GB | ~81 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
Fits only at 3-bit or lower
Too big for this machine
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), Qwen3.8-27B (17.9 GB), Granite 4.2 30B (17.9 GB), Jebadiah 27B (18.2 GB), Muse Glimmer 30B (18.3 GB), Mistral Small 22B (18.4 GB), AREX-2 27B (18.6 GB), Qwen3.6 27B (19.4 GB), Qwen3-Coder-30B-A3B (20 GB) and more.
Questions
What is the best AI model for the M2 (24 GB)?
gpt-oss-20b at MXFP4: the largest model on our shortlist of current models that runs at 4-bit or better at a usable speed (about 20 tok/s, ModelFit est.).
How much memory can a model use on the M2 (24 GB)?
About 16.0 GB of the 24 GB.
How fast is the M2 (24 GB) for local LLMs?
Decode speed is limited by memory bandwidth: 100 GB/s. A model that reads 10 GB per token runs at under 10 tokens per second.