Skip to content
localmodel.run

Text model · Llama 4

Llama 4 Scout: RAM and VRAM requirements

Llama 4 Scout needs about 64.2 GB to run at Q4_K_M (the Q4_K_M GGUF file is ~60.87 GB to download; KV cache and overhead add the rest), or about 110 GB at Q8_0. The lightest hardware that runs it is Apple M4 Max (128GB).

Llama 4 family · 109B params (Mixture-of-Experts: activates only 17B of 109B params per token, so generation is faster than the total size suggests) · released Apr 2025 · 1.7M Ollama pulls.

Shopping for hardware? See what runs Llama 4 Scout →

License Llama 4 Community · Conditional ↓ 483.8K/mo ♥ 1.3K on HuggingFace
Q4_K_M GGUF
60.87 GB
Q8_0 GGUF
106.67 GB
Memory @ Q4 (4k)
~64.2 GB
Context
128 k

Q4_K_M and Q8_0 are GGUF download (file) sizes. Memory @ Q4 includes the KV cache and overhead, so it runs larger than the file on disk.

Will it run on your device?

Llama 4 Scout runs on 4 of 40 tracked devices at Q4_K_M.

4 run well 0 tight fit 36 too small
Fit check Q4_K_M
No, not enough memory
needs 64.2 GB usable 10.5 GB
See the full breakdown

Memory breakdown

Weights (Q4_K_M)60.87 GB
+
KV cache (4k)2.5 GB
+
Overhead0.8 GB
=
Total64.2 GB

How context length changes it

4k context ~64.2 GB 32k context ~81.7 GB 128k context ~141.9 GB

Longer context grows the KV cache, which for Llama 4 Scout is sized by its 17B active params, not the full 109B. It needs ~64.2 GB at 4k and ~141.9 GB at 128k.

Benchmark scores

BFCL · tool use

#12 of 19 ranked · Overall Acc on the Berkeley Function-Calling Leaderboard.

28.1%

Sourced third-party benchmarks for the full-precision Llama 4 Scout, not a specific quant. See the full leaderboard.

Quantization sizes

GGUF quantson disk
Quantization Size on disk
Q2_K 45.6 GB est
Q3_K_M 53.3 GB est
Q4_K_M (default) 60.87 GB
Q5_K_M 77.7 GB est
Q6_K 89.4 GB est
Q8_0 106.67 GB
FP16 218 GB est

Lower quant = smaller and faster, slightly lower quality. Q4_K_M is the common default.

Run it

Ollama
$ ollama run llama4:scout
llama.cpp
$ llama-cli -hf unsloth/Llama-4-Scout-17B-16E-Instruct-GGUF:Q4_K_M
LM Studio
$ lms get unsloth/Llama-4-Scout-17B-16E-Instruct-GGUF

Which devices can run Llama 4 Scout?

Apple Silicon Macs

RAM-only laptops

iPhone & iPad

Android

NVIDIA GPUs

AMD GPUs

Same job, different size

Pick by what fits your memory: step down to free up VRAM, or step up if you have headroom.

Next size down GLM-4.5-Air 106B Next size up Command A 111B

Similar models

Head-to-head

FAQ

How much VRAM or RAM does Llama 4 Scout need?

At Q4_K_M, Llama 4 Scout needs about 64.2 GB (weights ~60.87 GB + KV cache + overhead) at a 4k context. At Q8_0 budget ~110 GB.

What is the Q4_K_M GGUF file size of Llama 4 Scout?

The Q4_K_M GGUF file is about 60.87 GB to download, and the Q8_0 GGUF is about 106.67 GB. That is the weights file on disk; to run it you also need room for the KV cache and overhead, so budget ~64.2 GB of memory at Q4_K_M.

Can Llama 4 Scout run on a laptop?

Llama 4 Scout is large; you need a high-memory Mac or multi-GPU setup at Q4_K_M.

Is Llama 4 Scout cheaper to run because it is a MoE model?

It is faster, not lighter. Llama 4 Scout activates only 17B of 109B params per token (so it runs quickly), but all experts must stay in memory, so it still needs memory for the full 109B.

Can I use Llama 4 Scout commercially?

Conditionally. Llama 4 Community License: free under 700M MAU, with use restrictions.

Understand the numbers

Short guides to the ideas behind Llama 4 Scout's memory and quant figures.

Llama 4 Scout, 109B MoE (17B active), natively multimodal with very long context. Q4_K_M ~61GB across a 2-file series. Size from the unsloth GGUF repo (Ollama llama4:scout is a comparable ~63GB build).

Sources

Last validated 2026-08-03. Memory figures are estimates. See methodology.