We're reducing our bandwidth prices.

Learn more
July 16, 2025
Early Access

Edge caching for web services entering early access

Coming to early access, paid web services will support edge caching for static content (text files, images, audio, and so on). By serving cached assets directly from Render's global CDN (similar to static sites), you can shorten your web service's response times and reduce load:

A diagram depicting edge caching for web services
A diagram depicting edge caching for web services

To apply for early access, please upvote the feature request at feedback.render.com. We'll gradually roll out early access to requesting workspaces over the coming weeks.

After early access is enabled for your workspace, you'll be able to enable edge caching for a web service from its Settings page in the Render Dashboard.

For preliminary details on edge caching, see the documentation.

July 15, 2025
Deprecated

Deprecated: Legacy service sharing

Some workspaces created before January 1, 2023 currently support sharing individual services with Render users outside the workspace:

The legacy sharing UI in the Render Dashboard (only visible to specific workspaces)
The legacy sharing UI in the Render Dashboard (only visible to specific workspaces)

This legacy sharing method is now deprecated. Future platform changes might disrupt access granted by this method. There is currently no scheduled date for end-of-life.

Collaboration remains central to the Render platform. If you currently use this sharing method, you can add affected users to your workspace and assign them appropriate roles.

July 14, 2025
Adjusted

Changed the default overlapping deploy policy for new workspaces

Starting 2025-07-14, newly created workspaces set their default overlapping deploy policy to Wait instead of Override. This policy determines how your services handle newly triggered deploys if a deploy is already in progress:

  • Wait: Allow the in-progress deploy to finish, then proceed directly to the most recently triggered deploy (skipping any in between).
  • Override: Immediately cancel the in-progress deploy and start the new one.

For most use cases, the Wait policy helps your workspace maintain a steady cadence of deploys during periods of high change volume.

Existing workspaces keep their current policy setting. Learn more about overlapping deploys in the documentation.

July 08, 2025
New

Create private link connections in the Render Dashboard

Professional workspaces and higher can now create private links directly in the Render Dashboard:

Creating a private link in the Render Dashboard
Creating a private link in the Render Dashboard

Private links enable your Render services to securely connect to AWS-hosted cloud services:

Diagram depicting private link communication
Diagram depicting private link communication

Use a private link to access compatible services like Snowflake, MongoDB Atlas, or resources in your own VPC. Each workspace can have up to 3 private links.

July 01, 2025
Adjusted

Upcoming changes to outbound bandwidth

On August 1, 2025, we're making the following changes to how we charge for bandwidth:

  • Lower bandwidth pricing: Bandwidth pricing will decrease from $30 per 100GB to $15 per 100 GB. We will continue to only charge for bandwidth usage above your workspace plan's included amount. If you don’t exceed the bandwidth already included in your plan, these changes won't affect your bill.
  • Accurate bandwidth billing: Currently, bandwidth usage only includes HTTP responses sent from your web services and static sites to browsers and other clients over the public Internet. We are correcting this behavior to align with industry standards. Outbound bandwidth will now include the following in addition to HTTP responses:
    • Service-initiated outbound traffic: Network communication initiated by any of your Render services over the public Internet (e.g., calls to an external database or uploads to a public cloud storage service)
    • Service-initiated Private Link traffic: Same as the above, but over a Private Link connection
    • WebSocket Responses: WebSocket responses sent from your web services to browsers and other clients over the public Internet

This update brings our pricing model in line with industry norms and helps us accurately reflect real infrastructure usage.

99% of workspaces will see lower or unchanged bandwidth costs.

Learn more about these changes in the documentation.

June 26, 2025
Early Access

Connect to the official Render MCP server over HTTP

The official Render MCP server is now hosted at the following URL:

https://mcp.render.com/mcp

This hosted server will automatically update over time to reflect the latest release. Like the underlying project, the hosted server is in early access.

Apps that support streamable HTTP communication with an MCP server (such as Cursor) can now connect directly to Render's hosted server instead of running a local version. See example configuration.

Note that the MCP server does not yet support OAuth, and clients authenticate via an Authorization header. Learn more in the documentation.

June 26, 2025
Improved

Added shortcuts to Render Dashboard search results

Search results in the Render Dashboard now include shortcuts to common settings:

Shortcut search results in the Render Dashboard
Shortcut search results in the Render Dashboard

Jump directly to your API keys, webhook settings, compliance docs, and more—all with a few keystrokes.

June 24, 2025
Adjusted

"Neutral" and "Skipped" GitHub checks no longer block auto-deploys

For services that set their auto-deploy behavior to After CI Checks Pass, the following GitHub check conclusions now all count as a passing check:

  • success
  • neutral
  • skipped

Previously, only success counted as a passing check. Conclusions besides those listed above continue to count as a failed check.

Learn more in the documentation.

June 16, 2025
Early Access

Official Render MCP server now in early access

Use the official Render MCP server to manage your Render infrastructure with natural language in AI-powered apps like Cursor, Windsurf, and Claude:

An MCP-powered prompt response in the Cursor IDE
An MCP-powered prompt response in the Cursor IDE

The MCP server runs on your development machine alongside your AI app. With each prompt, your app intelligently determines whether to use the Render MCP server to perform relevant actions using the Render API:

Diagram illustrating Cursor using a locally running Render MCP server
Diagram illustrating Cursor using a locally running Render MCP server

Supported actions currently include:

  • Creating web services, static sites, and datastores
  • Fetching and filtering service logs
  • Listing a service's deploy history
  • Querying a Render Postgres database

Get started with the MCP server in the documentation.

June 12, 2025
Improved

Updated default language versions for native runtimes

Newly created services on Render's native language runtimes now use the following updated language and library versions by default:

  • Python 3.13.4
  • Node.js 22.16.0
  • Ruby 3.4.4
  • Go 1.24.4
  • Elixir 1.18.4

With the exception of Go services, you can always specify a different language version.

Existing services keep their original default version to prevent breaking changes.