What's new

New Features

• Meet Surfy: Introducing our new mascot Surfy the capybara! Goggles on, vibes high, Surfy is ready to help while you surf.
• Introduce Skills and Skills Hub: Create, discover, and manage skills directly in the app settings. The Skills Hub provides a unified platform to browse popular skills and integrate them with SurfMind.
• MCP Integration: Connect to external data sources and services through Model Context Protocol (MCP) connectors. Enable integrations with tools like Notion, Jira, Asana, GitHub, and more to give the AI access to your workspace data and APIs.
• Add Import/Export settings and conversation history so you can bring your SurfMind setup across browsers

Updates

• LaTeX & Content Rendering: Add LaTeX math rendering in chat responses and fix KaTeX warnings for better scientific content display.
• Refined In-Page Experience: adding font size control, better handling of selection and quick-action scenarios, and improved mobile detection for touch devices.
• Extended Model Support: Support for latest models and improved handling of advanced reasoning features across different AI providers.

Bug Fixes

• Hotkey & Extension Polish: Fix quick chat login issues, enhance hotkey configuration with conflict detection, and improve content-script recovery on tab activation.
• Mobile & Safari Improvements: Better iPad Safari extension popover handling, improved touch device detection, and refined mobile UI layouts.

New Features

• Browser Action: Enable agentic browser action including opening page, scrolling, clicking and typing. The agent will ask for permission before taking action. Browser Action can be disabled in App Settings.
• App Settings: Add new app settings to enable/disable app features including browser control and in-page tools.
• SurfMind is now live on Firefox and iPad.

Updates

• Enhanced PDF Handling: Increase the file limit to 30MB and improved chat with PDF performance.
• Google Docs & Word Quick Chat: Enhance Google Docs and Microsoft Word selection handling for quick chats.
• Improved In-Page Chatbox: Enhance text interaction inside the chatbox with tooltips, cross-frame quick chats, and smarter copy/paste behaviors.

Bug Fixes

• API & Auth Stability: Resolve Claude API multi-turn response resets, fixed memory leaks, and improved session persistence in the authentication handler.
• Extension Polish: Fix conflicting stale states, cleaned up theme-updating flicker, and resolved multi-listener iframe issues.

New Features

• Token Statistics: Display response stats with token count tracking across different API providers
• Add fork functionality to chat messages
• Add search functionality to chat history

Updates

• Improve image and link display
• Disable thinking in quick chat
• Improve quick chat behavior in Google Docs
  • Activate via shortcut
  • Select and copy text, then activate quick chat with a shortcut to automatically include the selected text

Bug Fixes

• Fix images in PDF, Notion and Obsidian export
• Fix attachment handling in message conversion

New Features

• Chat export: Export chats to PDF, TXT, JSON, MD, Notion and Obsidian
• Writing preferences with formality options in quick chat
• Copy options for formatted/plain text

Updates

• Update shortcuts and shortcuts settings
• Localize shared links and website links

Bug Fixes

• Fix login timeout/logout issue
• Fix Custom Model config

New Features

• Web Search: AI can now search the web in real time, with inline citation badges, source previews on hover, and a dedicated Sources button for full references
• Thinking Mode: Enable deep reasoning with model-specific thinking effort settings; supported across Grok, Gemini, and Claude APIs

Updates

• Refactor message generation for reading and searching actions
• Grant persistent permissions for AI actions without repeated confirmation prompts
• Add agent tools for large attached files

Bug Fixes

• Fix streaming and open-tab issues
• Fix JSON parsing for partial/streaming responses
• Improve citation handling to avoid breaking markdown and code block rendering

New Features

• Support file attachment (pdf, csv, text, code)
• View attached file directly on browser side by side with the extension
• Enable instant in-page AI assistance in the file viewer

Updates

• Update context length estimation
• Use file inline instead of reference
• Add loading indicator for user login

Bug Fixes

• Fix surfmind extension window got added to selected tabs
• Fix user avatar issue
• Fix summarize issue and unknown active tab
• Fix various stability and performance improvements

New Features

• Bring your own models (BYOK)
  • Add Custom API support with dedicated configuration UI
  • Add preset list for popular LLM inference tools like Ollama, llama.cpp, LM Studio and vLLM
• Smarter in-page writing
  • Introduce the Input Focus Dot to trigger AI on input focus
  • Enhance quick actions, reply prompts, and multi-version responses for more control over edits and rewrites
• Add Apple In-App Purchase support
• Switch to Credits system

Updates

• Improve response parsing and streaming reliability
• Support thinking display for custom models
• Better replace/insert behavior in all input fields
• Implement auto-retry for blocked streaming requests
• Show regenerate button for failed responses

Bug Fixes

• Fix undefined model edge cases
• Make surfmind popup in Arc browser work again
• Various stability and performance improvements

New Features

• Dark mode
• Support Sign in with Apple

Updates

• Slash command improvements
  • Enable slash commands while editing an inline message
  • Dismiss the slash command menu with Esc
• Refactor login flow for better stability

Bug Fixes

• Fix Safari storage issue
• Fix Safari window “read tab” behavior
• Resolve maximum iteration processing errors
• Improve layout and interactions for touch devices

Updates

• Update and refine prompt messaging
• Improve survey experience
• Improve welcome section with more content
• Add insert and replace button for input elements

Bug Fixes

• Fix Continue in Chat state not persisting in the side panel
• Fix in-page issues (layout, positioning, size and spacing)

New Features

• Introduce in-page tool (Quick chat) for interacting with selected text directly on web pages
• Add quick prompts and slash (/) command support for faster actions
• Add user prompts, allowing users to create, save to quick prompts, and pin prompts for faster access
• Enable Add to Chat to seamlessly move selected text into the main chat
• Introduce Open Page tool to enable page-level actions

Updates

• Replace the extension popup with a new window so the extension can stay open independently
• Refactor authentication, popups, and internal structure for better reliability
• Improve streaming, keyboard navigation, and overall performance

Bug Fixes

• Resolve Safari and iOS issues including login, layout, and feature availability
• Fix window focus, tab selection, and multi-window behavior
• Prevent page shortcuts from interfering with chat input
• Fix translation, localization, and locale loading issues
• Fix various UI and styling issues across toolbar, sidebar, and chatbox

New Features

• Add OpenRouter support to access more models with capability-aware behavior
• Introduce System Prompt control, giving users more transparency and control over how models respond
• Enable chat sharing to share and manage chats easily with different sharing modes
• Enable rewards program to gain extra free credits

Updates

• Improve chat behavior based on selected model capabilities and context limits
• Add scroll to bottom button for long conversations
• Simplify and improve the model selector with provider and capability-based filtering

Bug Fixes

• Fix authentication, session sync, and token refresh issues
• Handle chat error more reliably
• Fix UI issues across popups, dialogs, history, and mobile views

New Features

• Introduce Google login with free credits to get started instantly
• Introduce subscription plans
• Add GPT-5.2 and Gemini 3 Flash models

Updates

• Refresh UI across popups, toggles, shortcuts, and the model selector
• Improve streaming response display for better readability and consistency

Bug Fixes

• Improve data storage stability and prevent sync loops
• Fix several UI and stability issues across the extension

Updates

• Add Claude Opus 4.5
• New keyboard shortcut (⌘E) to open the sidepanel and popup
• Better table and code formatting for clearer, cleaner outputs
• General storage updates for better reliability

Bug Fixes

• Fix issue with missing thought signature for Gemini 3 Pro

Updates

• Image uploads now always use JPEG for improved compatibility
• Add built-in image compression to reduce upload size and speed up performance

Bug Fixes

• Fix display issues with conversation items on phones for a smoother, more reliable mobile view
• Fix an issue where conversations couldn’t be deleted properly

New Features

• Support multi-tab chat
  • Improve behavior when reading pages, managing permissions, and syncing tabs
• Improve chat experience
  • Add message toolbar and enable inline edit
  • Enable regenerate for all asistant messages
  • Enable stop streaming
  • Smoother streaming with background processing enhancements
  • More reliable handling of AI responses formatting for code and tables
• New Models & Integrations
  • Add Gemini 3 Pro
  • Add GPT-5.1
  • Add Claude Haiku 4.5
  • Enhance LLM integration and improve tool-call handling

Bug Fixes

• Fix translation issues, language loading problems, and search bar in tab selector
• Fix bugs related to editing/regenerating messages when popup is closed
• Fix sync issues, read-page toggle bug, styles, and general UI inconsistencies

New Features

• Improve chat experience
  • Cleaner interface and refined styling
  • Faster, smoother scrolling with auto-scroll fixes
  • Send button now disables while responses are streaming
• Add 50+ languages
• Support image input
  • Upload, drag & drop, or paste images directly into the chat.
  • Enjoy smoother previews, better error handling, and more reliable image processing
• Support both page context and free-form chats with "Read Page" toggle
• New "Summarize" button

Updates

• Better API key setup flow
• General cleanup and performance enhancements

Bug Fixes

Lots of improvements across Safari, toggles, dialogs, sidepanels, image clearing, and more

Updates

• Add Claude Sonnet 4.5
• Add capture screenshot tool
• Improve OpenAI integration and fixed display issues for a smoother AI experience
• Show the popup when the side panel isn’t available, ensuring uninterrupted access
• Fix several UI and stability issues across the extension