Braintrust has released several new features this week, including improved experiment loading times and a new tutorial on finetuning GPT3.5 to write SQL queries. The company also evaluated the Alpaca evals leaderboard in Braintrust, providing aggregated rankings and individual model comparisons. Additionally, Datasets have been improved with display of last edited date and version number, while release notes highlight various bug fixes and enhancements. Braintrust is an enterprise-grade stack for building AI products, offering a range of tools to simplify the incorporation of AI into business operations.