Meta has published the model card and usage instructions for Llama-4-Maverick-17B-128E-Instruct, a multimodal large language model available under the Llama 4 Community License.
The documentation provides integration guides for libraries such as Transformers, inference providers including vLLM and SGLang, and local deployment via Docker Model Runner. The license agreement effective April 5, 2025, grants non-exclusive rights to use, reproduce, and modify the materials, requiring attribution and compliance with an Acceptable Use Policy.
This release enables developers to deploy and fine-tune the model using standard open-source tools while adhering to Meta's specific distribution terms.