github bytedance/UI-TARS-desktop v0.2.2

latest releases: @agent-tars@0.3.0-beta.9, @agent-tars@0.3.0-beta.5, @agent-tars@0.3.0-beta.4...
one month ago

Key Changes

  • support headful browser with VNC control
  • add model availability check logic

Details

VNC Browser

In this update, we have replaced the remote browser operator's screen casting feature with VNC Browser. This version provides a more stable screen casting experience and supports displaying the full Chrome UI:

vnc-browser-1080p.mp4

Check Model Availability

After configuring the VLM Model Settings, users can proactively click the Check Model Availability button below to verify the availability of the VLM Model:


What's Changed

  • chore(ui-tars): update release version by @ZhaoHeh in #824
  • chore(mcp-browser): add custom logger and addMiddleware by @ycjcl868 in #813
  • fix(ui-tars): action parser edge case action Chinese colon by @ycjcl868 in #825
  • docs(agent-tars): new home page by @ulivz in #841
  • docs: refine readme by @ulivz in #843
  • feat(ui-tars): add model availability check logic by @skychx in #894
  • feat(ui-tars): update volcano engine FaaS url by @skychx in #895
  • feat(ui-tars): update model check logic by @skychx in #899
  • feat(remote-browser): support headful browser with VNC control by @ZhaoHeh in #898

Full Changelog: v0.2.1...v0.2.2

Don't miss a new UI-TARS-desktop release

NewReleases is sending notifications on new releases.