Improvements

• Update support model list for Baidu Qianfan.

Bug Fixes

• Missing i18n resources.

Merged improvements and bug fixes from stable version.

• Update with new models for some LLM providers.
• Optimized performance and memory usage.
• Optimized performance of PlantUML editor

Improvements

• Improved overall performance of the CSV editor.
• Improved the performance of the Markdown editor.
• Auto-removes empty trailing columns before saving in the CSV editor.
• Updated support model list for almost all LLM providers.

Bug Fixes

• Fixed deleting cells in the first row not updating the table header in the CSV editor.
• Fixed paste functionality not working correctly when copied cells contain commas (,) in the CSV editor.
• Fixed inability to export images from PlantUML with transparent background — now uses PNG as export format.
• Fixed various missing internationalization resources.

Updated dependencies

Features

• Custom AI model provider(OpenAI Compatible).
• Support Baidu Qianfan AI models.

Improvements

• Agent with custom model provider.
• Allow to edit the name of an agent or a dataset.
• Add more models to Qianfan, Moonshot, ChatGLM.
• Remove deprecated models from DeepSeek.

Bug Fixes

• Max output tokens is incorrect for some Qwen models.
• Max output token is not applied to Moonshot api.
• Model provider setting check for Generate/Summarize is incorrect in some cases.

Update dependencies and refactor

Improvement:

• Add more models to Moonshot, ChatGLM.
• Remove deprecated models from DeepSeek, OpenAI.

Bug Fixes

• Automatic preview synchronization while editing Markdown files was not working correctly after updating to the latest version of JavaFX.
• Opening a Markdown file with scrolling content does not automatically synchronize with the preview.
• Missing i18n resources.

• Allow to change the orientation of the splitter of Mind Map editor.

• Hide floating information when the attribute view of the Mind Map editor is showing.
• Update i18n for the general preferences panel.
• Update models for Gemini and Qwen.
• Remove deprecating models from Qwen.

• Temperature parameters are restricted on the Kimi models.
• Incorrect max output token for Qwen model qwen-flash.

Improvements:

• Update i18n for AI generation and embedding.
• Update new models for OpenAI(gpt-5.5,gpt-5.5-pro) and Qwen(qwen3.6-27b etc), .
• Removed deprecated models for OpenAI, Qwen, Moonshot and ChatGLM.
• Add more information to metrics.

Bug Fixes:

• Checking update doesn't work well.
• Workaround for RichTextFX weird exceptions.

Update langchain4j

Improvements:

• Disable the thinking mode for AI models by default.
• Update Qwen model qwen3.6-flash and qwen3.6-plus.
• Update Moonshot model kimi-k2.6.
• Update DeepSeek models deepseek-v4-flash and deepseek-v4-pro.
• Update more i18n translations.

Bug Fixes:

• Test connection button should be disabled when no embedding store type is selected.

Update dependencies.

Improvements

• i18n refactor.
• Update GLM models.
• Add app launch statistics metrics.

Bug Fixes

• undo/redo doesn't work on note editor of mind map topic attribute view.

Update JavaFX to 25.0.3 and other dependencies.

Merge bug fixes from stable version.

Improvements:

• Use file name as default content in the first cell for a new created CSV file.

Bug Fixes:

• The model provider choice box should be disabled in the Generate&Summarize dialog when it is running.
• Language in internal embedding model details is incorrect.
• Failed to call some LLM apis after packaging to installer.
• Remove deprecated pre-defined models for Qwen.