What's New iOS 26 Liquid Glass • Beautiful new translucent design for iOS 26 • Full backward compatibility with older iOS versions • Applied to model selector, scroll button, and more Reasoning Models Support • New collapsible ThinkingView to see AI reasoning in real-time • Streaming support for thinking content • Perfect for Qwen3 and other reasoning-capable models Enhanced Markdown • Table rendering support • Clickable links • ~~Strikethrough~~ text support Changes • Updated MLX and dependencies to latest versions • 2025 AI models registry refresh Fixes • Improved thread card tappable area and preview text Thank you for using Eris!

What's New 2025 Models Update New state-of-the-art models: • Qwen3 0.6B & 1.7B - Multilingual with reasoning capabilities • Phi-4 Mini - Microsoft's most powerful compact model • Qwen Coder 0.5B & 3B - Enhanced code assistants Changes: • Qwen3 0.6B is now the default model • Models reorganized for better clarity • Optimized performance with MLX 2025 Thank you for using Eris!

iOS 26 Beta Compatibility Notice: We've added a helpful warning for users running iOS 26 beta. Since beta versions of iOS often experience higher device temperatures, we want to ensure you're aware this may lead to faster battery drain during extended use.  You can choose to accept this warning once or dismiss it permanently.

* Shortcuts & Siri Integration: Use voice commands to chat, summarize, translate, or generate text with new system-wide shortcuts. Supports multiple tones, model selection, and chat saving options — all with clear, helpful examples. * Streamlined Interface: Cleaner input area, larger buttons, rounded corners, and subtle background colors for a more refined experience in both light and dark mode. * Improved Model Selector: Moved to the navigation bar with a visual chevron indicator, haptic feedback, and smooth spring animations. * Pinned Chats: Pin up to 3 important threads with a visual indicator and animated transitions. Includes a limit alert and context menu actions. * Better Keyboard Handling: Automatically dismisses keyboard after sending a message and keeps it closed after AI replies. * New Documentation View: Access helpful sections on getting started, models, chat features, tips, and more — directly from Settings. * Model Compatibility & Error Handling: Removed unsupported models (e.g., Mistral 7B, StableCode 3B), added model validation, clearer error messages, and support for unsupported model detection. * UI Fixes and Enhancements: * Fixed SafariView presentation in CreditsView — thanks to Pavel Koupa! * Fixed button content shapes in Model Management and Settings — thanks again to Pavel Koupa! * General Improvements: Smoother performance, consistent design components, and better system stability.

What’s New in v1.5 The app now shows your exact device model (e.g. iPhone 15 Pro) Added bug report & feature request section, with links to GitHub and Discord Developer Mode toggle added to Preferences (optional tools for advanced users) Improved download system with retry logic and better error messages Wi-Fi is now clearly required only for first model download Significant improvements to stability, memory handling, and device compatibility Refactored and optimized model management: better info, warnings, and sorting iPhone 11 (base model) is no longer supported due to memory issues Many UI, performance, and reliability improvements under the hood

- Reordered parser conditions to check lists before bold text - Added processInlineMarkdown function to handle bold, italic and code formatting within list items - Now correctly renders lists with inline markdown like '- **Title**: Content' - Add regex caching for syntax highlighting with NSCache - Implement memory warning handler to clear caches and reduce GPU memory - Add scroll-to-bottom floating button with smooth animations - Improve auto-scroll behavior for new messages - Initialize MemoryManager on app startup for automatic memory handling - Optimize code syntax highlighting with cached regex patterns - Implement stop button to cancel text generation mid-stream - Add thread-safe CancellationToken class for cross-actor cancellation - Show loading spinner when model is being loaded into memory - Change stop button color from blue to red for better visibility - Add haptic feedback when pressing stop button - Differentiate between model loading and text generation states - Prevent stop action during model loading phase Languages added: - Modern languages: Dart, Lua, Elixir, Clojure, Zig, Nim, Julia - Enterprise: Groovy/Gradle, PowerShell, Visual Basic .NET - Scientific: Fortran, MATLAB (existing), R (existing) - Legacy: Pascal/Delphi - Web frameworks: Vue.js, React/JSX/TSX, Svelte, Angular - Markup/Config: HTML, CSS, YAML, JSON, Markdown - DevOps: Dockerfile, Terraform, GraphQL - Mobile: Objective-C, Kotlin (existing) - Blockchain: Solidity Performance improvements: - Reduced regex compilation overhead in code blocks - Better memory management under pressure - More responsive UI with optimized scroll handling