Release 2.0.0

2.0.0

Local OCR Improvements

• Native on-device inference: Local OCR now runs inference directly inside the app, making it a lot faster. Python is no longer required.
• Three different OCR modes: text, math, and mixed.

File to Text

• New File to Text tool: Convert PDFs and image files into text with local mixed OCR, including text and math formulas in one output. This uses the new and improved Local OCR pipeline.
• Copy, export, and history: Converted files can be copied, saved as .txt, and reopened from the new File to Text history.

Chat and LaTeX Rendering

• LaTeX rendering in chat: Assistant replies with math now render directly in the chat instead of showing raw LaTeX only.
• Improved LaTeX previews: Snip history now renders better math thumbnails and cleaner preview text. Previews now render a lot faster.

Screenshot Capture

• Multi-monitor support: Screenshot and 2LaTeX capture now work across multiple displays with improved overlay handling.
• More reliable capture flow: The selection overlay is better organized, tracks the cursor across screens, and exits more cleanly after capture or cancel.