No — not on this device
Llama 4 Scout 109B-A17B at Q4_K_M needs 63.4 GB against 10.7 GB usable, and the shortfall of 52.7 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.
A 10M-token window on paper, chunked attention in practice (8K chunks on 3 of 4 layers). Needs 64 GB+ at Q4.
What hardware do I need for Llama 4 Scout 109B-A17B? →
Fits instead: Llama 3.1 8B Instruct (6.1 GB) · Llama 3.2 3B Instruct (3.3 GB)
The VRAM budget
Quantisation ladder
| Quant | Weights | Total @ 8K | Max context | Tok/s | Quality | Fit |
|---|---|---|---|---|---|---|
| Q8_0 | 107.9 GB | 110.0 GB | — | ~1.1 | −0.1% ppl | 99.3 GB over |
| Q6_K | 83.2 GB | 85.3 GB | — | ~1.4 | −0.4% ppl | 74.6 GB over |
| Q5_K_M | 71.9 GB | 74.0 GB | — | ~1.6 | −0.8% ppl | 63.3 GB over |
| Q4_K_M | 61.3 GB | 63.4 GB | — | ~1.9 | −1.9% ppl | 52.7 GB over |
| Q3_K_M | 49.6 GB | 51.7 GB | — | ~2.3 | −5.4% ppl | 41.0 GB over |
| Q2_K | 42.5 GB | 44.6 GB | — | ~2.7 | −15% ppl | 33.9 GB over |
Quality is the published perplexity delta against f16 weights. Max context assumes an f16 KV cache; q8_0 roughly doubles it. This model interleaves sliding-window layers (8192 tokens, 12 of 48 layers global), which is why its cache barely grows with context.
How to run it
$ pip install mlx-lm $ mlx_lm.generate --model mlx-community/Llama-4-Scout-17B-16E-Instruct-4bit \ --max-tokens 512 --prompt "Hello"
Apple's own array framework. The fastest path on Apple Silicon. More on MLX.