New Models at Oumi

Oumi added two new models to the platform: Qwen 3.8 27B and Muse Glimmer 30B.

Both models are available across the same workflow. Teams can evaluate them, generate targeted data, fine-tune the models, and deploy the results without moving between separate systems.

One workflow from evaluation to deployment

Start by benchmarking a model against the task. Generate data that targets the gaps, fine-tune the model, and deploy the result when it meets the bar.

This workflow gives teams a direct path from a model choice to a production-ready specialized system.