Changelog

  1. 1.15.0

    Current

  2. 1.14.0

  3. 1.13.0

  4. 1.12.0

  5. 1.11.0

  6. 1.10.0

  7. 1.9.0

  8. 1.8.0

  9. 1.7.0

    • New
      Added encryption for cloud sync: turn it on and everything this device sends to your Google Drive is readable only with a key your own browser holds, which Polymeti never receives and cannot reset.
    • New
      Added the recovery key that opens your encrypted data on another device, which can be copied, downloaded as a text file, or saved as an image to print or scan.
    • New
      Added an optional passphrase as a second way to unlock your synced data, for when typing the key is impractical.
    • New
      Added live voice conversation, so you can talk to a model and hear it answer, with the transcript written to the chat as you go.
    • Improved
      Improved the Cloud sync settings, where encryption has its own section and the list of what syncs is always in view.

  10. 1.6.0

    • New
      Added new models:
    • New
      Added web search and image generation switches to the retry menu.
    • Improved
      Made UI and UX improvements across the app, on the list pages, the model picker, the model pages and the chat screen.
    • Improved
      Improved the mobile version, with larger text and tap targets.
    • Fixed
      Fixed image generation on Grok models and API key sync between devices.

  11. 1.5.0

    • New
      Added GPT-6 Astra from OpenAI, which reads images and PDFs and lets you set how hard it thinks, from low up to max.
    • New
      Added Gemini 3.8 Flash from Google, which reads images, audio, video and PDFs, and lets you set how hard it thinks, from low up to high.
    • New
      Added support for HEIC and HEIF photos on Gemini models, so a picture straight from a phone can be attached without converting it first.
    • New
      Added a new home page layout, with the sidebar and the message box on it from the first visit.
    • Improved
      Made UI and UX improvements across the app, on the message box, the model picker and the public pages.
    • Improved
      Improved the mobile version, so the controls and the pages fit a phone screen.

  12. 1.4.0

    • New
      Added new models:
    • New
      Added support for local AI models running entirely on your device with no API key required:
    • New
      Added message queueing, so a follow-up can be typed while an answer is still arriving.
    • New
      Added a context window gauge with on-hover token counts below the message box.
    • New
      Added quick-toggle buttons for web search and image generation below the message box.
    • New
      Redesigned the AI model picker, with filters at the top and model capabilities (images, search, reasoning) on every row.
    • New
      Redesigned the Message Details window for better clarity.
    • New
      Added a prompt to reload a local model with more context when a conversation no longer fits.
    • Improved
      Improved the Explore page layout with capability indicators for each model's modalities and features.
    • Improved
      Expanded each model's page with its creator, the providers serving it, and what it is suited to.
    • Improved
      Optimized local models to download automatically before sending your message.
    • Fixed
      Fixed a bug causing low token counts for Google reasoning models and cached Claude prompts.

  13. 1.3.0

    • New
      Added an Explore page to browse supported models and check API key compatibility.
    • New
      Added bulk actions to archive, delete, or categorize multiple chats at once.
    • New
      Added this "What's New" page to track version updates and changes.
    • Improved
      Improved answer regeneration to automatically retain your original prompt settings.
    • Improved
      Moved DeepSeek V4 Pro from preview to general availability.

  14. 1.2.0

  15. 1.1.0

  16. 1.0.0

    MVP