Released: Jun 18, 2026

26.1.3 中的更新

特性

Rich Comments

• Mermaid Diagrams
  • Rich comments can now render Mermaid diagrams directly within source code comments, making it easy to document architecture, workflows, and designs without leaving your editor. Diagrams can also be generated automatically with AiGen and remain version-controlled alongside your source code.
• LaTeX Formulas
  • LaTeX formula support has graduated from beta with improved rendering quality and performance.

Embedded Content Features

• Thumbnail Mode
  • Right-click an embedded image or diagram and select Use Thumbnail to replace it with a compact, clickable thumbnail that expands to the full-size version when selected. This is ideal for keeping large visual assets from overwhelming the code editor.
• Fit in TextView
  • Right-click an embedded image or diagram and select Fit in TextView to automatically scale it to fit the width and height of the active TextView.
• Version Control Friendly
  • Embedded images, Mermaid diagrams, and LaTeX formulas are stored in your solution folder and participate naturally in version control.

New AI Model Integrations

• Expanded AI Model Support
  • AiGen now supports a wider range of industry-leading AI providers and models:
    • Anthropic Claude 4.5 model family, including extended thinking and configurable reasoning budgets.
    • Google Gemini 2.0, 2.5, and 3.1 model families.
    • OpenAI GPT-5.5 model family.
• Persistent AI Coding Preferences
  • AiGen can now remember your coding preferences across sessions.
• New AI Models Options Page
  • A new AI Models options page centralizes model configuration and dynamically adapts its interface to the selected model. Controls such as Temperature, Reasoning Effort, Thinking Budget, and AI Mode automatically appear only when supported by the selected AI model.
• Azure Endpoint Support
  • AiGen now supports Azure-hosted AI endpoints with context-specific endpoint selection, allowing powerful models to be used for reasoning and lightweight models to be used for context acquisition.
• MCP Server Integration
  • AiGen can now automatically integrate with both the DevExpress MCP Server and Microsoft MCP Server, enabling access to authoritative product documentation and API information while generating responses.
• AI Cost Tracking
  • AI cost tracking and cost estimates have been removed due to the increasing complexity and variability of pricing models across AI providers.

Performance & Intelligence Enhancements

• Surgical AI Code Editing - AiGen now performs smaller, surgical code edits that target:
  • Member signatures.
  • Method bodies.
  • Attributes.
  • XML documentation comments.
  • Individual code blocks within large methods.
  • Embedded mermaid diagrams/formulas.
• Smarter Context Packing
  • AiGen now gathers, prioritizes, and packs relevant project context more efficiently before sending requests to reasoning models, improving both response quality and token utilization.

Additional AI Improvements

• Diagrams, Formulas, and Code Working Together
  • AiGen now understands relationships between code, Mermaid diagrams, and LaTeX formulas.
• Clipboard Integration
  • You can now explicitly include clipboard contents in AI prompts or ask AiGen to place generated content directly onto the Windows clipboard from anywhere within Visual Studio.

Test Runner

• Improved AI Workflow Support
  • The Test Runner now includes a context menu command that copies reports of failing test cases directly to the clipboard (contained within the selected test nodes), making it easy to submit failures to AI for analysis.
• MSTest v4 Support
  • Added support for MSTest v4.
• Debugging Enhancements
  • AiGen can now analyze active compiler errors, call stacks, and runtime values to assist with live debugging.

User Experience Improvements

• Voice Interaction
  • AiGen now supports voice interaction with speech recognition and spoken feedback in more than 100 languages, making AI assistance accessible through natural conversation.
• Personalized AI Experience
  • AiGen now remembers your preferred coding style and conventions, allowing future interactions to automatically align with your established development preferences.