Llama 4 Scout
Meta
🧠 Language Model
Fast multimodal model. 109B params (17B active, 16-expert MoE). Vision, tool use, 327K context. Fits on a single GPU.
Specifications
Context Window327,680 tokens
Max Output8,192 tokens
Speed●●●●● Very Fast
ModalitiesInput: text, image · Output: text
FeaturesVision: Yes Tools: Yes Streaming: Yes
Pricing
Included with plan