No — not on this device
Llama 3.3 70B Instruct at Q4_K_M needs 42.8 GB against 27.0 GB usable, and the shortfall of 15.8 GB is more than 32 GB of system RAM can cover at a tolerable speed. A smaller sibling or a lower quantisation is the honest answer here.
Does not fit
8K context
Q4_K_M · 39.7 GB
Llama 3.3 Community
Released Dec 2024
Still the creative-writing favourite: consistent voice, takes direction. Needs 48 GB to sit comfortably on GPU at Q4.
What hardware do I need for Llama 3.3 70B Instruct? →
Fits instead: Llama 3.1 8B Instruct (6.1 GB) · Llama 3.2 3B Instruct (3.3 GB)
The VRAM budget
weights 39.7 GB
Weights 39.7 GB
KV cache @ 8K 2.50 GB
Runtime overhead 0.6 GB
Over budget 15.8 GB past 27.0 GB
Quantisation ladder
| Quant | Weights | Total @ 8K | Max context | Tok/s | Quality | Fit |
|---|---|---|---|---|---|---|
| Q8_0 | 69.9 GB | 73.0 GB | — | ~1.2 | −0.1% ppl | 46.0 GB over |
| Q6_K | 53.9 GB | 57.0 GB | — | ~1.6 | −0.4% ppl | 30.0 GB over |
| Q5_K_M | 46.6 GB | 49.7 GB | — | ~1.8 | −0.8% ppl | 22.7 GB over |
| Q4_K_M | 39.7 GB | 42.8 GB | — | ~2.1 | −1.9% ppl | 15.8 GB over |
| Q3_K_M | 32.1 GB | 35.2 GB | — | ~2.6 | −5.4% ppl | 8.2 GB over |
| Q2_K | 27.5 GB | 30.6 GB | — | ~3.0 | −15% ppl | 3.6 GB over |
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/Llama-3.3-70B-Instruct-4bit \ --max-tokens 512 --prompt "Hello"
Apple's own array framework. The fastest path on Apple Silicon. More on MLX.
01Download is 39.7 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 27.0 GB of 36 GB. Raising it with iogpu.wired_limit_mb is possible, and risky.
03If it falls back to CPU silently, drop the context first, then step down a quantisation.