LLMs · Meta

Llama 4 Scout 17B-16E

Name: Llama 4 Scout 17B-16E
Author: Meta

Parameters: 109B (17.0B active)
Architecture: Mixture of experts
Max context: 10,240K tokens
License: llama4
Released: 2025-04

How to run it

Quickest · Ollama

One command downloads a ready-to-run version and starts a chat.

ollama run llama4:scout

Get Ollama at ollama.com/download (opens in a new tab) · model page (opens in a new tab)

Quantized files · GGUF

Works in LM Studio, llama.cpp, Jan and Ollama. No sign-in needed.

huggingface.co/ggml-org/Llama-4-Scout-17B-16E-Instruct-GGUF (opens in a new tab)

ollama run hf.co/ggml-org/Llama-4-Scout-17B-16E-Instruct-GGUF

Original weights · Hugging Face

Full-precision files from Meta, for fine-tuning, vLLM, Transformers.

huggingface.co/meta-llama/Llama-4-Scout-17B-16E-Instruct (opens in a new tab)

Free Hugging Face account needed: request access on the model page, and Meta approves it (usually within minutes to a few hours).

Official repo · GitHub

Reference code, inference scripts and docs from Meta. The weights themselves download from Hugging Face.

github.com/meta-llama/llama-models (opens in a new tab)

git clone https://github.com/meta-llama/llama-models

Meta’s official Llama repo: model cards, prompt formats and download tools.

Links checked 5 days ago against Hugging Face, Ollama and GitHub.

Memory needed

Weights plus KV cache and runtime overhead. Leave a little extra free for the operating system and other apps.

Quantization	Quality	Weights	Total at 8K context	Total at 32K context	Download
`Q2_K`	Low	43.4 GB	45.7 GB	50.5 GB	GGUF files (opens in a new tab)
`Q3_K_M`	Low	53.2 GB	55.4 GB	60.3 GB	GGUF files (opens in a new tab)
`Q4_K_M`	Good	66.2 GB	68.5 GB	73.3 GB	GGUF · part 1 of 2 (opens in a new tab)
`Q5_K_M`	High	77.1 GB	79.3 GB	84.1 GB	GGUF files (opens in a new tab)
`Q6_K`	High	89.1 GB	91.3 GB	96.1 GB	GGUF files (opens in a new tab)
`Q8_0`	Very high	115 GB	117 GB	122 GB
`F16`	Full	217 GB	219 GB	224 GB	Original weights (opens in a new tab)

Best hardware for Llama 4 Scout 17B-16E

The least expensive GPU in our catalog that runs it without offloading is the NVIDIA RTX PRO 6000 Blackwell (launch price $8,565) at Q4_K_M.

Grade · points	Hardware	Quant	Memory	Speed	Fit
85 ptsExcellent (S) · 85 of 100 points: Model quality 30.8/45, Memory fit 25.0/25, Speed 15.0/15, Memory use 15.0/15	NVIDIA RTX PRO 6000 Blackwell	`Q4_K_M`	73.3 GB / 96.0 GB	64.7 tok/s	Runs well
80 ptsGreat (A) · 80 of 100 points: Model quality 30.8/45, Memory fit 25.0/25, Speed 9.8/15, Memory use 15.0/15	Apple M4 Max (128 GB)	`Q4_K_M`	76.1 GB / 96.0 GB	17.0 tok/s	Runs well
80 ptsGreat (A) · 80 of 100 points: Model quality 30.8/45, Memory fit 25.0/25, Speed 11.8/15, Memory use 12.8/15	Apple M2 Ultra (192 GB)	`Q4_K_M`	80.0 GB / 144 GB	24.9 tok/s	Runs well
78 ptsGreat (A) · 78 of 100 points: Model quality 30.8/45, Memory fit 25.0/25, Speed 7.4/15, Memory use 15.0/15	NVIDIA GB10 (128 GB)	`Q4_K_M`	77.7 GB / 115 GB	9.9 tok/s	Runs well
77 ptsGreat (A) · 77 of 100 points: Model quality 30.8/45, Memory fit 25.0/25, Speed 12.0/15, Memory use 10.1/15	Apple M3 Ultra (256 GB)	`Q4_K_M`	83.8 GB / 192 GB	25.5 tok/s	Runs well
77 ptsGreat (A) · 77 of 100 points: Model quality 30.8/45, Memory fit 25.0/25, Speed 6.7/15, Memory use 15.0/15	AMD Ryzen AI Max+ 395 (128 GB)	`Q4_K_M`	76.1 GB / 96.0 GB	8.0 tok/s	Runs well
73 ptsGreat (A) · 73 of 100 points: Model quality 30.8/45, Memory fit 25.0/25, Speed 12.0/15, Memory use 6.0/15	Apple M3 Ultra (512 GB)	`Q4_K_M`	99.2 GB / 384 GB	25.5 tok/s	Runs well
59 ptsGood (B) · 59 of 100 points: Model quality 22.7/45, Memory fit 10.0/25, Speed 11.5/15, Memory use 15.0/15	Apple M4 Max (64 GB)	`Q2_K`	49.5 GB / 48.0 GB	23.7 tok/s	Runs with offload
57 ptsUsable (C) · 57 of 100 points: Model quality 22.7/45, Memory fit 10.0/25, Speed 9.9/15, Memory use 15.0/15	Apple M1 Max (64 GB)	`Q2_K`	49.5 GB / 48.0 GB	17.4 tok/s	Runs with offload
49 ptsUsable (C) · 49 of 100 points: Model quality 22.7/45, Memory fit 4.0/25, Speed 7.7/15, Memory use 15.0/15	NVIDIA GeForce RTX 5090	`Q2_K`	47.3 GB / 32.0 GB	10.5 tok/s	Heavy offload

Check it on your own hardware

Estimates come from published hardware specs and our open formulas, not from running the model on your machine. Real speed varies with runtime, drivers and settings. Numbers marked “Measured” come from a linked benchmark.