Yes — with 141.4 GB to spare
Llama 4 Scout 109B-A17B at Q4_K_M fits your CPU only · DDR5 8-channel server entirely on the GPU at 8K context, at an estimated 6.0 tokens per second. There is room for its full 1024K window.
Fully on GPU
8K context
Q4_K_M · 61.3 GB
Llama 4 Community
Released 5 Apr 2025
Vision
A 10M-token window on paper, chunked attention in practice (8K chunks on 3 of 4 layers). Needs 64 GB+ at Q4.
The VRAM budget
weights 61.3 GB
Weights 61.3 GB
KV cache @ 8K 1.50 GB
Runtime overhead 0.6 GB
Free 141.4 GB of 204.8 GB
Quantisation ladder
| Quant | Weights | Total @ 8K | Max context | Tok/s | Quality | Fit |
|---|---|---|---|---|---|---|
| Q8_0 | 107.9 GB | 110.0 GB | 1024K | 3.4 | −0.1% ppl | Long context |
| Q6_K | 83.2 GB | 85.3 GB | 1024K | 4.4 | −0.4% ppl | Long context |
| Q5_K_M | 71.9 GB | 74.0 GB | 1024K | 5.1 | −0.8% ppl | Long context |
| Q4_K_M | 61.3 GB | 63.4 GB | 1024K | 6.0 | −1.9% ppl | Recommended |
| Q3_K_M | 49.6 GB | 51.7 GB | 1024K | 7.4 | −5.4% ppl | Long context |
| Q2_K | 42.5 GB | 44.6 GB | 1024K | 8.6 | −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. 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
$ llama-server \
-hf meta-llama/Llama-4-Scout-17B-16E-Instruct:Q4_K_M \
-c 8192 -ngl 99
The engine underneath most of the others. Every knob is exposed. More on llama.cpp.
01Download is 61.3 GB. Keep it on an SSD — a first load off a spinning disk takes minutes.
02With no GPU, thread count matters more than clock. Start at one thread per physical core.
03There is room to go to the model's full 1024K context on this card.