Llama 4 Scout Instruct

Meta llm

Meta's first natively multimodal Llama, a 17B-active-parameter mixture-of-experts model (16 experts, 109B total) with a 10M-token context window.

overall ⌄

34.1

81 of 93 ranked

price · $/M tokens

self-hosted

open weights, no first-party API

to run it yourself ⌄

65 GB

minimum · 262 GB recommended

Benchmarks

What its maker published, and what anyone else measured. Every figure links the document it came from.

No recorded lineage — root or standalone model.

org	Meta
released	2025-04-05
params	109B-A17B
license	llama-4
hf id	meta-llama/Llama-4-Scout-17B-16E-Instruct