This release improves overall reliability and performance by enhancing how we handle timeouts for data warehouse and Redis connection failures. It also introduces new APIs for managing user and dataset permissions—both at the account level and across the enterprise.
This release focuses on stability improvements and bug fixes related to account management and knowledge ingestion.
OpenAI/Azure GPT‑4.1 is now the default model everywhere—including customer VPCs. This build also introduces widget branding hooks and further enhances our advanced‑reasoning and MCP‑workflow betas.
This release improves post table summaries in chat experiences and introduces a beta of advanced reasoning within chat experiences which would eventually replace the classic chat experiences as we know it. Finally, this release also enhances workflows with deep research nodes.
This release enables enterprise API keys which can be used to perform account management action, a directed-graph workflow engine & UI, first class support for Elastic Search within VPCs, MS SQL Server dataset-copy/update operations, and queries where 2 tables within a dataset may not join along with well over a hundred stability fixes.
192.0.0 contains several meaningful improvements to deep research and workflows.
In addition, we have several important bug fixes across the product, but particularly with dataset jobs.
In this release, we introduce new knowledge-layer APIs, allowing users to create and update dimensions and metrics programmatically in their datasets. We also continue to develop our beta deep research and workflow planning features. We have added a beta connector for Microsoft SQL Server. Finally, we have made numerous performance optimizations that help the application feel more responsive.
This release introduces dataset tagging using #, deep research agent in beta, and the ability to bulk update sensitivity for dimensions and metrics.
It also includes several bug fixes.
This release resolves several issues related to copy and update jobs and introduces a new bulk update feature for dimension and metric sensitivity.
This is a feature-filled and exciting release for enterprise users. In particular, this release includes our first version of our enterprise admin console, which allows enterprise admins to manage datasets in base account and distribute changes to these datasets to their downstream customers in an organized way. For more details, please visit our freshly published Enterprise Administration Guides. As this is our first version, we know there is room for improvement, so we are looking forward to your feedback.
We also include the ability to "reset" feedback chats and have removed all v1 functionality.
As usual, we also have a large number of bugfixes and performance improvements.