Baseten Training - Model training built for production inference.

When we announced the beta launch of Baseten Training on May 19, our goals were two-fold: 1) To bring Baseten's performant infrastructure to AI model training. 2) To address core pain points our customers were reporting to us when training models — limited access to compute, opaque pricing, underutilized capacity, and limited control over what's running under the hood. Since then, we’ve worked hand in hand with early users who have run thousands of training jobs, ranging from small LoRA-supervised finetunes to multi-node RL runs. After months of working closely with early users and receiving overwhelmingly positive feedback, we're excited to announce that Baseten Training is now generally available (GA) to all businesses on Baseten. Training delivers tools such as caching, checkpointing, deploy-from-checkpoint, and new training recipes, making it the most agile training infrastructure for developers. Try it today and let us know what you think!