Oumi Updates: CLI and New Model

The latest Oumi updates make it easier to work from the terminal, improve models in production, and run new inference workloads.

GPT-6 Astra for inference

GPT-6 Astra is now available for inference use cases on the platform, including evaluation, synthesis, model-as-judge, and failure-modes analysis.

Continuous improvement with monitoring

Monitoring supports a full self-improvement loop for models in production. Teams can connect production signals to the next round of evaluation and improvement.

Autoscaling deployments

One-click deployment now supports autoscaling. Ship a model in a single step, then let the deployment scale down when it is idle.

Command-line access

The Oumi CLI brings platform workflows into the terminal. Teams can work with Oumi from scripts and developer tools, not only from the browser.