Every memory option compared, 273 to 307 GB/s memory bandwidth. Updated 2 November 2026.
M4 Pro / M5 Pro are sold with 24, 48, 64 GB of memory. Counting the 31 models tracked here: 24 GB runs 23 models well; 48 GB runs 27 models well; 64 GB runs 27 models well. Memory bandwidth is 273 to 307 GB/s, about 2.3 times the memory speed of a base M4 (120 GB/s), so the same model generates words proportionally faster. Apple Silicon shares one pool of memory between the processor and the GPU, and macOS lets the GPU use about two thirds of it on Macs up to 36 GB and about three quarters above that, so the memory you buy decides which models load at all.
| Model | 24 GB | 48 GB | 64 GB |
|---|---|---|---|
| Llama 3.3 70B | No | No | Slow 3/s |
| Wan 2.2 14B | No | Great about 14 min | Great about 14 min |
| LTX-2.3 | Slow about 4 min | Great about 2 min | Great about 2 min |
| Qwen 3.6 35B A3B | No | Great 41/s | Great 41/s |
| Qwen 3.5 35B A3B | No | Great 43/s | Great 43/s |
| Gemma 4 31B | Tight 7/s | Good 8/s | Good 8/s |
| DeepSeek R1 32B | Tight 7/s | Good 8/s | Good 8/s |
| Qwen3 Coder 30B | Tight 35/s | Great 39/s | Great 39/s |
| Qwen 3.6 27B | Tight 7/s | Good 8/s | Good 8/s |
| Gemma 4 26B A4B | Tight 28/s | Great 31/s | Great 31/s |
| Z-Image Turbo 6B | Great about 27 s | Great about 27 s | Great about 27 s |
| Wan 2.2 5B | Great about 7 min | Great about 7 min | Great about 7 min |
| gpt-oss 20B | Great 32/s | Great 32/s | Great 32/s |
| ACE-Step 1.5 | Great about 1 min | Great about 1 min | Great about 1 min |
| FLUX.2 klein 4B | Great about 19 s | Great about 19 s | Great about 19 s |
| Phi-4 14B | Great 16/s | Great 16/s | Great 16/s |
| Gemma 4 12B | Great 18/s | Great 18/s | Great 18/s |
| Stable Diffusion XL | Great about 11 s | Great about 11 s | Great about 11 s |
| Qwen3 TTS 1.7B | Great about 30 s | Great about 30 s | Great about 30 s |
| Gemma 4 E4B | Great 45/s | Great 45/s | Great 45/s |
| Qwen 3.5 9B | Great 22/s | Great 22/s | Great 22/s |
| Qwen3 VL 8B | Great 23/s | Great 23/s | Great 23/s |
| DeepSeek R1 8B | Great 28/s | Great 28/s | Great 28/s |
| Qwen2.5 Coder 7B | Great 31/s | Great 31/s | Great 31/s |
| Gemma 4 E2B | Great 72/s | Great 72/s | Great 72/s |
| Whisper large-v3 turbo | Great about 42 s | Great about 42 s | Great about 42 s |
| Qwen 3.5 4B | Great 44/s | Great 44/s | Great 44/s |
| Kokoro TTS | Great about 6 s | Great about 6 s | Great about 6 s |
Text models show words a second. Image, video and voice models show the time per job (image, 4 second clip, minute of speech or similar).
As an Amazon Associate I earn from qualifying purchases.
About 23 GB is usable by the GPU. The largest chat or coding model that fits is Gemma 4 31B (21.5 GB), generating about 7 words a second with 1.3 GB to spare. The quickest is Gemma 4 E2B at about 72 words a second. Closest miss: Qwen 3.5 35B A3B, which needs about 23.6 GB (1 GB more than this configuration can give the GPU).
Find M5 Pro Macs with 24 GB on Amazon
About 46 GB is usable by the GPU. The largest chat or coding model that fits is Qwen 3.6 35B A3B (24.6 GB), generating about 41 words a second with 21 GB to spare. The quickest is Gemma 4 E2B at about 72 words a second. Closest miss: Llama 3.3 70B, which needs about 45.6 GB (1 GB more than this configuration can give the GPU).
Find M5 Pro Macs with 48 GB on Amazon
About 61 GB is usable by the GPU. The largest chat or coding model that fits is Qwen 3.6 35B A3B (24.6 GB), generating about 41 words a second with 36.2 GB to spare. The quickest is Gemma 4 E2B at about 72 words a second. Closest miss: gpt-oss 120B, which needs about 68.8 GB (8 GB more than this configuration can give the GPU).
For everyday chat, 24 GB runs 23 of 31 models well. 48 GB to 64 GB unlocks nothing new in this list. The extra memory only gives longer conversations and more room for other apps.
Yes, from 24 GB of memory, at about 7 words a second.
24 GB runs gpt-oss 20B at about 32 words a second; 48 GB runs gpt-oss 20B at about 32 words a second; 64 GB runs gpt-oss 20B at about 32 words a second.
As an Amazon Associate I earn from qualifying purchases.
Unlocks 3 more: Llama 3.3 70B, Qwen 3.5 122B A10B, gpt-oss 120B.