together-dedicated-containers

Deploy custom Docker containers for GPU inference on Together AI using the Jig CLI to build, push, and run images. It streamlines the container lifecycle from local build to production deployment on managed GPU infrastructure without managing hardware.

The Queue API handles asynchronous job submission for inference workloads, supporting prioritized and scalable tasks. It enables progress tracking for custom Dockerized pipelines running on managed GPU infrastructure.

Yes, you can run multimedia generation pipelines like image and video processing on dedicated containers. The setup supports GPU-accelerated workloads beyond standard model endpoints using the Sprocket SDK for request handling.

Together AI provides private container registry access for hosting custom model servers with image versioning and security controls. This allows teams to securely store and manage Dockerized inference workloads before deployment.

Handle inference requests inside a Dockerized model server using the Sprocket SDK. It provides a consistent in-container workflow for processing custom GPU inference tasks submitted through the Queue API.

Use dedicated GPU containers instead of standard model endpoints when running specialized models, custom pipelines, or multimedia generation requiring asynchronous queuing. They are necessary for workloads needing private image hosting and custom orchestration.