Llama 4 Maverick
Meta
🧠 Language Model
Most capable open-weight model. 400B params (17B active, 128-expert MoE). Multimodal with vision, tool use, and 524K context.
Specifications
Context Window524,288 tokens
Max Output8,192 tokens
Speed●●●●● Fast
ModalitiesInput: text, image · Output: text
FeaturesVision: Yes Tools: Yes Streaming: Yes
Pricing
Included with plan