Releasesby GlyphReduced median first-build time for Docker-based previews by 45%
Docker-based service previews and preview environments now use the build cache from their base service’s most recent successful build. Previously, each preview’s first build started with an empty cache.
Also shipped Nov 11, 2026Render in November 2026 →
Web search in apps
Retool added native web search capabilities to the app builder Chat tab and app functions, enabling agent applications to query live web data via a managed Tavily account.
Eval datasets, curated by Bayes from your agents' real conversations
Monte Carlo introduced automated evaluation dataset generation by Bayes, creating Golden and Improvement datasets from an agent's scored conversations across intent and issue clusters.
Pages are generally available
Databricks announced the general availability of Pages in Unity Catalog, providing governed definitions of business concepts organized by domain on the Discover page.
BigQuery project caps are generally available
Project caps, also known as scheduling policies, are now generally available for BigQuery Enterprise and Enterprise Plus editions. This feature allows users to limit maximum slots and concurrency per project within a reservation.
BigQuery reservation assignments support principal routing
BigQuery now allows specifying a principal property on reservation assignments to route queries to specific reservations based on the identity of the user or service account. This feature is generally available for Enterprise and Enterprise Plus editions.
BigQuery JSON_FLATTEN function is generally available
The JSON_FLATTEN function is now generally available. It extracts non-array values from JSON inputs, including those nested within consecutive arrays.
More from RenderEvery Render release →
Render Sandboxes now available in beta
Now in public beta, use Render Sandboxes to spin up isolated, ephemeral compute environments for executing untrusted code, such as agent-generated apps and scripts. Create and manage sandboxes programmatically using the Render CLI or the TypeScript and Python SDKs.
Upcoming service preview and preview environment support for GitHub stacked PRs
On November 18, 2026 , Render will begin supporting separate service previews and preview environments for each pull request in a GitHub PR stack: Currently, Render creates previews only for PRs that directly target your service's linked branch. This means that only the bottom PR in a stack can generate previews.
Updated default versions of Node.js and Bun
New services on Render now default to Node.js 24.22.1 and Bun 1.4.4.
Added support for Bun canary builds
Render now supports canary builds for Bun. Canary builds let you preview upcoming Bun features before they reach a stable release. To use canary builds, do one of the following: Set your service's BUN_VERSION environment variable to canary . Add a .bun-version file to the root of your repo with the single line: canary .
Added two Key Value plans and lowered prices for larger existing plans
Render now supports two additional compute plans for Key Value instances: 3 GB RAM with up to 3,000 simultaneous connections 50 GB RAM with up to 50,000 simultaneous connections We've also reduced pricing for existing Key Value compute plans that provide at least 1 GB RAM.
Added Blueprint support for Render Workflows
You can now manage your workflow services using Blueprints , Render's infrastructure-as-code model. Define workflows alongside other service types in your Blueprint YAML.
See your service's deploys on a new page
Services now have a new Deploys page in the Render Dashboard. The Deploys page features: Your service's deploy history, without a date cutoff. A Live badge indicating the current deploy. Status of in-progress deploys. A banner flagging when auto-deploys are disabled.
Introduced the Flex compute plan for Render Workflows
Render Workflows uses a new default compute plan for task runs: flex . The flex compute plan provides up to 1 CPU and 4 GB of RAM. Task runs on flex are billed only for the CPU and RAM they actually use, resulting in significantly lower costs for computationally inexpensive runs.