Skip to content

Release Notes

Generate AI PC v2.4.6 - 3/11/2025

Overview

This release introduces workflow improvements, LLM integration updates, and system optimizations to enhance the Generate AI PC experience. Users now have better control over document processing, authentication, and performance settings.

New Features & Enhancements

Export & Import Cards

  • Users can export workflow cards as a ZIP file.
  • Directory Sync for Cards: Drop the uncompressed folder into:
    ~/Documents/Generate/imported_cards/
  • Click the Sync button in the Card Manager to load the new cards.

Import LLMs & Directory Sync

  • Supports GGUF models for Nvidia, AMD, and Mac.
  • Supports OpenVINO LLM for Intel GPUs.
  • Directory Sync for LLMs: Drop the model file into the appropriate folder: 
Drop GGUF / OpenVINO LLM folder
  • Click Sync in the LLM Manager to load the model.

Document Settings & Precision Controls

New document extraction precision settings allow users to fine-tune how documents are processed:

  • Standard Mode: Very fast, optimized for text-only documents.
    Standard Mode

  • High Precision Mode: Best for tables and structured documents.
    High Precision

NPU for Vector Database Creation

  • Users must download models before enabling NPU functionality.
  • Corruption Warning: If LLM inference is running, corruption may occur. Restarting the application is required.
  • Recommended Setup:
  • For Nvidia/AMD GPUs: Use CPU for embedding.
  • For Intel CPUs: Enable NPU acceleration.

LLM Query Enhancements

  • Users can now add additional instructions to LLM queries.
    High Precision
  • Multi-File Selection for QA allows querying across multiple files.
  • Selected Field Rendering now displays numbered output.

Authentication & Subscription Management

  • New email-based login verification to improve security.
  • Multiple device support: Users can log in from multiple devices with the same email.
  • Subscription management features:
  • Add and manage license keys.
  • Individual subscription tracking.

Code & UI Enhancements

  • Rendering improvements for code snippets:
  • Python
  • C++
  • JSON data types
  • Tooltips added for action buttons across the UI.
  • RAG Power Introduced: Enhancing retrieval-augmented generation capabilities.

Performance Improvements

  • Optimized app start time to reduce initial loading delays.

Enjoy the new features, and let us know your feedback!