Llama 4 Scout VRAM and RAM requirements

Mixture-of-experts. All 109B weights load; 17B are active per token. The 10M context is not covered by the 20% overhead.

Total weights

109B

Active

17B

Context

10M tokens

License

Llama 4 Community

meta-llama/Llama-4-Scout-17B-16E-Instruct on Hugging Face · 162,682 downloads on the Hub snapshot in this catalog

Memory

• Q4

  65.4 GB

• Q8

  115 GB

• FP16

  261 GB

• GGUF file size from the linked Hub repo. KV cache was not added, because this model's attention shape is not in the catalog.
• Formula estimate: parameter count × bytes per weight × 1.20. No GGUF file size is stored for this quant.

This model advertises 10M tokens. Filling that window needs more KV cache than the 20% allowance.

Compatible hardware

8 of 64 machines fit at Q4 or better. Fits means the estimate is inside the usable memory. Tight means less than 2 GB of room.

Amazon links are affiliate searches, not live prices.