What’s new
in Molar.

Updates to the browser, the test environment, and the way you work with Molar.

Website

Clearer pricing, security and setup pages.

The website now states the beta terms and the security controls in plain words, and every developer page shows commands you can copy and run.

  • Pricing lists the beta limits: 25 hosted runs and $10 of AI model use a day per organization, no card.
  • The security page covers where runs can go, what the AI model sees, access controls and the static IP for allowlisting.
  • The subprocessors list names Azure OpenAI as the AI model provider and no longer lists Vercel.
  • Developer tools shows the CLI install and MCP setup; Browser agents shows an HTTP API example.
Browser agent

A new browser agent for hosted runs.

Hosted task runs use a new browser agent. It checks each step against your goal as it works and can export the run as a Playwright test.

  • Saved logins and protected values are typed only when needed, on your project’s hosts.
  • Checks run where your task says “verify,” and appear in the exported test at the same point.
  • Runs are named after their goal, and recordings show the cursor.
  • The agent recovers when a page crashes or a tab closes, instead of failing the run.

Available on hosted task runs.

Platform

AI model requests moved to Azure OpenAI.

Hosted runs now use a model deployed on Microsoft Azure OpenAI Service in India. The subprocessors list was updated on October 26.

  • Model requests no longer go to Anthropic.
Legal

Privacy policy, terms, DPA and subprocessors.

Molar published its privacy policy, terms of service, a data processing addendum template, a subprocessors list and a security page, and made the beta free.

  • Free during the beta, with daily limits per organization.
  • A DPA template you can ask to sign.
Browser execution

Checks for each requested browser step.

Molar checks each requested step before marking a run complete. A successful final screen no longer hides a missed action earlier in the flow.

  • Clearer results when a browser task stops before completing every step.
  • More responsive cancellation when you stop a run.
  • More consistent use of saved test context on repeat runs.

Available in the browser-testing beta.

Workspace

One dashboard for the Molar products.

Cartographer, Clones, Guard, and Trace share the Molar dashboard, making it easier to move between browser work, test environments, and run evidence.

  • Shared product navigation and organization settings.
  • A common entry point for product workflows.

Feature availability varies during beta.

Put the next update
to work.

Create a free workspace and describe the first flow to test.

Free during the beta. No card.