Skip to content
mlx.app

M2 Ultra · 128 GB

800 GB/s memory bandwidth · 108.8 GB usable for models · 19.2 GB left to macOS

The short answer

37 models run comfortably and 0 more will run tight at 8K context. The largest that fits is GPT-OSS 120B at q4, generating around 229 tokens per second.

macOS19.2 GBweights65.7 GBKV cache0.6 GBheadroom42.5 GB

Every model on this Mac

ModelSizeBest quantVerdictSpeed est.
Qwen2.5 0.5B Instruct490Mbf16FITS · 107.7 GB FREE669 tok/s
BGE-M3567Mbf16FITS · 107.7 GB FREE578 tok/s
Qwen3 0.6B600Mbf16FITS · 106.7 GB FREE547 tok/s
Parakeet TDT 0.6B v2600Mbf16FITS · 107.6 GB FREE547 tok/s
Llama 3.2 1B Instruct1.2Bbf16FITS · 106.1 GB FREE265 tok/s
Whisper Large v31.6Bbf16FITS · 105.7 GB FREE212 tok/s
Qwen3 1.7B1.7Bbf16FITS · 104.5 GB FREE193 tok/s
Llama 3.2 3B Instruct3.2Bbf16FITS · 101.9 GB FREE102 tok/s
Qwen3 4B4Bq8FITS · 103.3 GB FREE154 tok/s
Qwen3 Embedding 4B4Bq8FITS · 103.3 GB FREE154 tok/s
Mistral 7B Instruct v0.37.3Bq8FITS · 100.0 GB FREE84.6 tok/s
Qwen2.5 7B Instruct7.6Bbf16FITS · 93.1 GB FREE43.2 tok/s
InternVL3 8B8Bq8FITS · 100.3 GB FREE77.2 tok/s
Llama 3.1 8B Instruct8.0Bbf16FITS · 91.7 GB FREE40.8 tok/s
Qwen3 8B8.2Bq8FITS · 98.9 GB FREE75.3 tok/s
Qwen2.5-VL 7B Instruct8.3Bq8FITS · 99.5 GB FREE74.4 tok/s
Gemma 2 9B IT9.2Bq8FITS · 99.0 GB FREE66.8 tok/s
Llama 3.2 11B Vision Instruct10.7Bbf16FITS · 87.4 GB FREE30.7 tok/s
Qwen2.5 14B Instruct14.7Bq8FITS · 91.6 GB FREE42.0 tok/s
Qwen3 14B14.8Bq8FITS · 91.7 GB FREE41.7 tok/s
GPT-OSS 20B21Bq8FITS · 86.1 GB FREE172 tok/s
Codestral 22B v0.122.2Bq8FITS · 83.3 GB FREE27.8 tok/s
Mistral Small 24B Instruct 250124Bq8FITS · 82.0 GB FREE25.7 tok/s
Gemma 3 27B IT27Bbf16FITS · 54.8 GB FREE12.1 tok/s
Gemma 2 27B IT27.2Bq8FITS · 79.9 GB FREE22.7 tok/s
Qwen3 30B-A3B Instruct 250730.5Bq8FITS · 75.6 GB FREE187 tok/s
Qwen3 Coder 30B-A3B Instruct30.5Bq8FITS · 75.6 GB FREE187 tok/s
Qwen2.5 32B Instruct32.5Bq8FITS · 72.1 GB FREE19.0 tok/s
Qwen2.5 Coder 32B Instruct32.5Bq8FITS · 72.1 GB FREE19.0 tok/s
DeepSeek R1 Distill Qwen 32B32.5Bq8FITS · 72.1 GB FREE19.0 tok/s
Qwen3 32B32.8Bq8FITS · 71.8 GB FREE18.8 tok/s
Qwen2.5-VL 32B Instruct33Bq8FITS · 71.6 GB FREE18.7 tok/s
Mixtral 8x7B Instruct v0.146.7Bq8FITS · 58.1 GB FREE47.9 tok/s
Llama 3.3 70B Instruct70.6Bq8FITS · 31.1 GB FREE8.7 tok/s
Qwen2.5 72B Instruct72.7Bq8FITS · 28.9 GB FREE8.5 tok/s
Llama 3.2 90B Vision Instruct88Bq4FITS · 59.3 GB FREE13.3 tok/s
GPT-OSS 120B116.8Bq4FITS · 42.5 GB FREE229 tok/s
Qwen3 235B-A22B235BOVER
DeepSeek V3671BOVER