Snowflake announced the public preview of Cortex Training, its serverless offering for customizing and deploying open-weight LLMs. According to the article, every production AI application generates millions of valuable data points about a business, including customer preferences, usage patterns and outcomes, but most enterprises never take advantage of it, and many default to building agents on closed foundation models that can be costly and lack detailed understanding of the business. With Cortex Training, teams can use familiar training frameworks or build advanced custom training loops with low-level serverless APIs without managing GPU provisioning, scheduling, checkpointing and distributed execution. The offering is built on a highly customizable and optimized open source backend, bringing the flexibility and efficiency of a complex, self-managed bare-metal training cluster to a simplified serverless experience. Benchmarks show up to 61% lower cost than competing serverless offerings. Teams can use Snowflake’s integrated data platform to curate training-ready data sets and capture agent traces and logs, creating a feedback loop for model customization and improvement. The article states post-training methods such as SFT and RL are how LLMs learn what good behavior looks like and how to make more accurate, cost-effective decisions using an organization’s specific context.