Deploy Any Hugging Face Model Instantly with Goose and Together DCI
Deploying machine learning models often involves navigating a maze of setup complexity: configuring inference servers, setting up container environments, and understanding model-specific requirements. Together.ai is aiming to eliminate those barriers with its Dedicated Container Inference (DCI) platform, allowing developers to deploy any Hugging Face model in production-ready GPU environments with minimal effort. The process leverages Goose, a command-line interface (CLI) agent runner, alongside Togethers DCI infrastructure. The result? A seamless deployment experience that skips the usual setup headaches. How it Works Consider Netflix‘s recently released Void-Model, which removes objects from videos while accounting for their interactions with the environment. Traditionally, deploying such a model would require days of setup. With Together’s tools, developer Blaine Kasten was able to deploy it on release day in just three steps:Install the Together DCI skill: Using the command npx skills add togethercomputer/skills, Goose gains the ability to configure Togethers infrastructure for any model.Run a single command: A simple prompt like I want to deploy this model on Togethers dedicated containers https://huggingface.co/netflix/void-model initiates the entire deployment process.Let the agent handle the rest: Goose automatically configures the inference server, generates container files, and deploys the model, producing a working setup hosted on Together infrastructure. The output of this process was