Product versions

This page provides a list of AI Assistant versions and their corresponding release notes.

252.26199.202

Claude Agent, powered by Anthropic’s Agent SDK, is now available in the AI Chat.
Track quota consumption and Top-up AI Credits directly in the AI widget.

The Next Edit Suggestions feature is now in preview, offering real-time code modification recommendations as you type.

Project rules: Guide AI with coding standards and tech stack, shared across your team.
New file types supported for code completion: The list of supported file types now includes SQL, YAML, JSON, Markdown, and more.
Local multiline completion is now available for Java and C++.
Choose your preferred local models for code completion: Connect models like Qwen2.5, Codestral, or the open-source Mellum.
Context improvements: Reference the project structure and attach images in the chat.
Auto-trim large prompts to optimize token usage and cloud quota.
Updated model selector: AI Assistant automatically picks the best model for chat-based tasks and shows visual cues for cost, reasoning strength, and experimental status.
Card verification for the Free tier: This is required for some AI Free and Trial users to prevent abuse (no charges).

New LLM in chat: Claude 4 Sonnet.
Edit mode Enhancements: AI-powered suggestions now offer improved context handling and seamless project integration.
Pre-Commit Checks: Added AI-driven pre-commit support with suggestions and one-click Quick Fix options.
Improved Code Completion: Improved context gathering and added better filter models with critical fixes preventing completion from being shown.
Expanded Language Support: Embedding-based indexing extended to WebStorm, PhpStorm, and other IDEs.
Smart Project-Wide Search: AI-powered searching for feature discovery and advanced code analysis is now available.
Enhanced Codebase Mode: AI Assistant now uses chat history to generate more contextually relevant responses.
Database Schema Integration: Added the ability to attach/detach database schemas through the AI Assistant.
Improved PyCharm Actions Visibility: PyCharm-specific AI actions in DataWrangler are now easier to access.
Local Embedding Support for C/C++: Extended local embedding functionality to support additional use cases for C/C++.
Commit Tool Enhancements: Improved handling of resolved and conflicted files during non-modal merges.

New LLMs in chat:
Removed from the model selector in chat:
- Gemini 1.5 Flash
- Gemini 1.5 Pro
- o1-mini

New LLM in chat: Gemini 2.5 Pro.
New subscription tiers: AI Free, AI Pro, AI Ultimate.
Offline mode: Use local models for AI-powered help without an internet connection.
Multi-file edits: Apply changes across multiple files directly from the chat using advanced RAG.
Web search from chat: Use /web to fetch documentation and answers directly from the internet.
Apply snippets from chat: Instantly insert suggested code into the right spot in your project.
MCP server integration: Connect custom tools to AI Assistant via the Model Context Protocol.
Restrict AI usage with .aiignore: Control where AI Assistant can operate in your codebase.

AI chat model selection: Choose from multiple AI models, including Google Gemini, OpenAI, Anthropic, or your own local model via Ollama or LM Studio.
Advanced code completion with broader language support: Cloud code completion now supports more languages, including JavaScript, TypeScript, HTML, C#, C++, Go, PHP, Ruby, and Scala.
Inline AI prompts: Enter your prompts directly in the editor, and AI Assistant will recognize it and generate a suggestion.
Context management in chat: View and manage every element included as context.
Integrated IDE Documentation: Access your IDE's documentation directly within the AI chat.
AI-powered quick fixes: Accelerate error resolution with intelligent, context-aware quick fixes.
File-wide code generation: Streamline edits across an entire file for Java and Kotlin.

23 October 2025