Cline v4.1.11 Changelog

What's new

• AI Image Generation: If the AI model you are using supports it, it can now create pictures as part of its task. These images will show up right inside your chat conversation.
• See What You're Paying: The app will now show you the actual billed cost when you use the Cline gateway (the service that connects you to the AI models).

Improvements

• Bigger Model Menu: The list of AI models you can choose from got a major refresh. It adds a bunch of new providers (like AMD, RunInfra, and others) and updates the pricing and default models for all of them.
• Better Provider Links: When you need to sign up for an AI provider, the app now sends you straight to the page where you get your API key, instead of a generic homepage.
• Smarter Chat Filters: The app will no longer confuse you by offering image, voice, or other non-chat models in the menu when you are just trying to pick a text model to chat with.
• Clearer AI Activity: When using the Claude Code provider, you can now see all the behind-the-scenes tools the AI runs, instead of those actions being hidden from you.

Fixes

• VS Code Command Bug: Fixed an issue where code actions would fail with a "command not found" error on the newest version of VS Code (version 1.134).
• File Mention Spaces: Fixed a bug where using the @ symbol to mention a file wouldn't work if the file's folder path had spaces in it.
• Multi-line Edit View: Fixed a glitch so you can properly see the "diff" view (comparing the old code to the new code) when editing multiple lines in files that use CRLF line endings (a Windows-specific way of ending text lines).
• Smoother Task Resuming: When you resume a paused task, the app now picks up the surviving session exactly as it was, instead of trying to rebuild it from scratch and potentially losing your progress.
• No Settings Bleed-Through: Fixed a problem where custom settings for one task would "leak" into your next task. Now, when you switch or clear tasks, those temporary settings are wiped clean.
• LiteLLM Token Limits: Fixed an issue where the app was accidentally overwriting your custom input limits (how much text the AI can read at once) for LiteLLM with default catalog values.
• Gemini Custom URLs: Restored the ability to use custom web addresses for Gemini, and fixed old host-root settings so they keep working.
• Fixed Duplicate Commands: Fixed an issue where loading skill commands accidentally pasted their instructions twice into your message.
• Restored Hook Functions: Fixed a bug where tool hooks (custom automated actions) weren't sending their context changes to the AI, and made sure the app waits for finishing hooks so their output and cancel buttons work right.
• PowerShell Fail-Fast: Fixed a bug where running PowerShell commands on Windows would keep trying to run even after an error happened, flooding you with output and falsely claiming it succeeded. Now it stops at the first error.
• Legacy Bundle Context Errors: Fixed an older version of the app so it only tries to "recover" from a context overflow (when an AI gets too much text) if the error message specifically says so. It will no longer trigger this recovery mode for unrelated errors, like a bad tool schema or an oversized image.
• Other Tech Fixes: Fixed minor bugs causing structured commands and remote configurations to crash or behave unexpectedly.

Breaking changes

No breaking changes in this release.