Yes — with 87.3 GB to spare
Mistral NeMo 12B at Q4_K_M fits your M4 Max · 128 GB entirely in unified memory at 8K context, at an estimated 44 tokens per second. There is room for its full 128K window.
Fully in unified memory
8K context
Q4_K_M · 6.9 GB
Apache 2.0
Released Jul 2024
Multilingual 12B with a 128K window, built with NVIDIA. A roleplay and fiction favourite that refuses to die.
The VRAM budget
weights 6.9 GB
Weights 6.9 GB
KV cache @ 8K 1.25 GB
Runtime overhead 0.6 GB
Free 87.3 GB of 96.0 GB
Quantisation ladder
| Quant | Weights | Total @ 8K | Max context | Tok/s | Quality | Fit |
|---|---|---|---|---|---|---|
| F16 | 22.7 GB | 24.6 GB | 128K | 13 | Reference | Long context |
| Q8_0 | 12.1 GB | 13.9 GB | 128K | 25 | −0.1% ppl | Long context |
| Q6_K | 9.3 GB | 11.2 GB | 128K | 33 | −0.4% ppl | Long context |
| Q5_K_M | 8.1 GB | 9.9 GB | 128K | 38 | −0.8% ppl | Long context |
| Q4_K_M | 6.9 GB | 8.7 GB | 128K | 44 | −1.9% ppl | Recommended |
| Q3_K_M | 5.6 GB | 7.4 GB | 128K | 55 | −5.4% ppl | Long context |
| Q2_K | 4.8 GB | 6.6 GB | 128K | 64 | −15% ppl | Long context |
Quality is the published perplexity delta against f16 weights. Max context assumes an f16 KV cache; q8_0 roughly doubles it.
How to run it
$ pip install mlx-lm $ mlx_lm.generate --model mlx-community/Mistral-Nemo-Instruct-2407-4bit \ --max-tokens 512 --prompt "Hello"
Apple's own array framework. The fastest path on Apple Silicon. More on MLX.
01Download is 6.9 GB. Keep it on an SSD — a first load off a spinning disk takes minutes.
02macOS caps what the GPU may wire down at about 96.0 GB of 128 GB. Raising it with iogpu.wired_limit_mb is possible, and risky.
03There is room to go to the model's full 128K context on this card.