Releases: iOfficeAI/OfficeAI
Releases · iOfficeAI/OfficeAI
V0.6.0
V0.6.0
General Updates
- Optimized the typesetting/layout solution to fix issues where template styles were lost or failed to refresh
- Added import and export support for layout templates
- Added support for translating content in multiple ways, including normal paragraphs, tables/cells, selected text, and mixed content
- Fixed detection and repair for the “Disable all plugins” option switch
- Fixed an issue where the custom prompt input box could not scroll when the content was too long
- Automatically corrected line break issues (inconsistent line breaks between web-copied content and documents)
- Fixed a crash issue when the Office/WPS language version was not Simplified Chinese, Traditional Chinese, or English
- Fixed an issue where the API key could not be used in translation scenarios
- Added a points/credits management feature to the Settings window
v0.5.8
V0.5.8
General Updates
- The writing intelligent agent locally supports adding, modifying, and deleting custom templates
Optimizes response performance in redline/greenline mode, with speed comparable to revision mode
Translation feature supports preserving original text (i.e., side-by-side translation)
Proofreading feature supports retaining revision marks
Optimizes performance during multi-round fragmentation
*Fixes issues with Chinese-English punctuation conversion
v0.5.7
General Updates
- In the proofreading scenario, users can now choose whether to display differences using red/green highlights or Track Changes mode.
- In the default chat scenario, model replies now output the full text instead of partial content (only plain text during generation, with full rendering at the end).
- Optimized output rendering performance for proofreading, polishing, formatting, default chat, and writing scenarios.
- Fixed an issue where multiple panels could appear after repeatedly opening functions in the Office environment.
- Fixed a slowdown issue when closing a document without performing any actions after opening it.
- In the formatting scenario, moved Times New Roman to the top of the Western font selection list.
- In translation, proofreading, polishing, and continuation scenarios, if the confirmation dialog appears but the user switches to another function without taking action, a prompt now reminds the user to complete the current action first.
- Restored Excel formula functionality.
- Fixed an issue where the model occasionally returned Markdown tags in proofreading and polishing scenarios.
- Added a desktop repair tool to allow users to quickly fix issues when problems occur.
- Fixed improper handling of
<br>tags returned by the model in polishing and proofreading scenarios. - Fixed an issue where multiple settings windows could appear in specific versions of WPS.
- Fixed specific formatting bugs occurring under WPS.
- Redesigned the application icon.
v0.5.6
- Add a custom navigation bar feature in Word
- Optimize the polishing and proofreading interaction interface
- Enhance layout to support long text
- Support translating only selected parts in Word tables
- Improve the wording of the Chinese single/double quotation mark conversion tool in the toolbox
- Fix inappropriate wording after executing “AutoFit table to contents” in the Word toolbox
- Fix the bug in the page number display logic of the layout feature
- Fix the issue where clicking the layout style edit window caused it to lose focus or be blocked
- Display page number position settings in the layout feature using a multilingual scheme
- Fix incorrect display when deleting default field template settings
- Fix the issue where field variable paragraph formatting did not take effect in WPS
V0.5.5
V0.5.5
General Updates
- Support polishing for long texts and tables
- Support image recognition feature for large models
- Fix issue where translation feature fails to apply to multi-paragraph content
- Fix issue where proofreading feature fails to apply in certain scenarios
- Fix issues with formatting layout
- Fix issue where HTTP/HTTPS prefixes are lost when saving API key service providers
V0.5.4
General Updates
- Support long-text processing for proofreading, polishing, and translation functions
- Optimize typesetting code and reintroduce one-step undo tool (implemented as a two-step restoration solution)
- Improve slow paragraph traversal issue in Office applications
- UI optimization for custom mode: add examples, include word limit prompts, and add public superscripts
- Integrate content summarization, meeting minutes, and weekly report assistant into a new summary function
- Add "Add New Provider" entry to service provider list in API key mode
- Automatically correct URL when model list refresh fails in custom (OpenAI protocol) under API key mode
- Fix issue where clicking "Extract/Filter" in Excel toolbox triggers quick image insertion function
- Resolve misalignment issue in Excel's name alignment feature
- Fix unrecoverable number/currency format brush issue in Excel
- Distinguish between window-adaptive and content-adaptive table processing in Excel toolbox
- Adjust style window to fix decimal point input failure
- Fix WPS ghost character deletion issue before paragraphs
- Correct multilingual loading and English display adaptation issues in style panel
- Optimize knowledge base selection interaction scheme
- Fix false repair exception report when only WPS is installed
- Use COM interface to check availability when both Office and WPS are installed
- Enhance Office installation status detection
v0.5.3.1
- Adds AI Navigation page, supporting custom addition of navigation pages
- Adds support for personal edition knowledge base
- Qwen3 adds support for optional inference modes
- Adds support for Western fonts in typesetting
- Adds support for outline directory settings and application in typesetting
- Fixes the issue where residual VBA code prevents the next task from executing
v0.4.0
New Features:
✔ Enabled Excel deep integration - Directly analyze and compute table data
✔ Added HTML analysis results display within plugin with PDF export support
✔ Implemented batch table insertion in Excel
✔ Added local model support for VLLM
Bug Fixes:
🔧 Fixed unresponsive clicks in Office/WPS on some machines
🔧 Resized WPS plugin interface that was too small to adjust
🔧 Fixed missing menu issue in some Office installations
🔧 Resolved Word 2010 compatibility issues
🔧 Optimized performance for long documents (polishing, proofreading, formatting)
🔧 Fixed "Continue Writing" feature failure in Word
🔧 Stabilized intermittent formatting function issues
OfficeAI 0.3.20
- Support for left-right linkage in custom API/local mode
- Support for custom addition, deletion, and modification of API-KEY models and addresses
- Enhanced configuration repair functionality, including checking data directory permissions
- Fixed the issue where the Excel extraction and filtering function caused the program to freeze when selecting an entire column
- Fixed the issue where the program failed to start in environments with Green Shield encryption software
- Fixed the program crash caused by administrator permission issues
- Fixed the issue "save message to db failed." due to permission problems
- Fixed the issue where the lack of system patches on Windows 7 prevented installation

OfficeAI v0.3.16 supports Deepseek-R1
- Local deployment supports APIKEY mode for Alibaba and Tencent DEEPSEEK-R1
- Added custom streaming conversation duration
- Corrected some interaction copy
- Optimized the user experience of DeepseekR1
- Fixed an issue where the local server did not start in API-KEY mode, causing errors
- Fixed an issue where vendor errors were not displayed in API-KEY mode
- Optimized some prompt copy
- Supported the thinking chain UI display for Deepseek-R1
- Official website, API-KEY, and local deployment all support DeepseekR1
- API-KEY supports more Deepseek mirror sites
- Optimized copy for slow vendor server
- One-click formatting function supports the layout of red-headed documents
- One-click formatting supports custom addition of layout rules and more rule options
- Some interaction optimizations on the local server side