• Work on multiple conversations independently, with automatic context summaries for long chats. • Improved sidebar gestures and message actions, with a shortcut to the latest message. • Clearer image generation feedback, in-chat audio playback, and native video previews. • Improved startup responsiveness, recovery after background interruptions, and iPad keyboard, larger text, and window layouts.

• Meet Wanta: a redesigned native chat workspace for conversations, attachments, and generated content. • Find chats in the sidebar with search, favorites, and custom titles. Older on-device history remains available as a read-only archive. • View tool steps, retry interrupted attachment uploads, and open saved images, audio, and video. • Improved chat scrolling, reply cancellation, and recovery after returning to the app.

• Generated image results recover more reliably after reopening a conversation, returning to the app, or reconnecting • Image previews now ignore stale background loads when switching messages or closing preview, reducing cases where an older image replaces the current one

• Saving extensionless images from preview to Photos is now more reliable • Very long user messages open in a collapsed preview first, making conversations faster to enter and scroll • Generated image conversations load more smoothly by moving tool result and cached image processing off the main screen path • Camera opening and conversation list loading are more stable in edge cases

• Attachment upload failures now show clearer recovery guidance instead of a generic request failure • Files without extensions upload more reliably by detecting common image, document, and media types from file contents • Zoomed image previews now pan more naturally, reducing conflicts with horizontal page swipes • Expired or timed-out image generation results now stop loading and show the correct failed state

• Chat lists scroll more smoothly with better layout caching for long messages, AI replies, and reasoning content • Generated image browsing is more stable, with less repeated rendering and background preview decoding • Thinking states update more calmly, reducing repeated animations and layout refreshes while waiting for replies

• Send button states are clearer, making idle, ready-to-send, and sending states easier to tell apart • Generated image results are more stable, with fewer repeated placeholders or incomplete result lists after completion • Conversation titles are more accurate when replies mainly come from tool results, reducing empty or off-topic titles

• Login is more complete, with SMS verification and WeChat login entry points plus smoother email and phone code flows • Message retry now refills the original prompt more accurately and can follow remote configuration for automatic composer append behavior • Image uploads and generated image results are more reliable, with oversized images compressed before sending and repeated generated results shown correctly • Upgrade prompts and connection state handling are more dependable, with remote minimum-version configuration and fewer false disconnects after late timeouts

• Starting the app now opens a fresh chat state more reliably, while reusing an empty draft when possible • Long user messages can be collapsed and expanded, and retry actions stay anchored to the right prompt • Image preview opening transitions are smoother, reducing jumps when moving from chat into full-screen preview • Upgrade prompts are clearer when the current client version is no longer supported

• Image previews are smoother, with improved large-image decoding and steadier system preview transitions • Async image edit results are more reliable, with less repeated polling noise when temporary server errors occur • Voice input cancellation is more consistent, including the image editing flow • Generated image preview coverage has been expanded so future image browsing changes are easier to verify

• Added an inline email sign-in flow and refined the Google sign-in entry for a smoother login experience • Invitation rewards now support shareable posters with invite codes and multiple visual styles • Voice input is steadier, with persisted input mode preference, first-recording haptics, and polished recognition progress feedback • Long toast messages can now wrap, and debug reporting is more tightly scoped for clearer diagnostics

• Image results returned by generic connectors are recognized more reliably during chat sync and replay • Image uploads, remote image previews, and attachment size handling are more stable • The chat composer now uses the system camera picker for a steadier photo capture flow • Chat UI details have been refined, including user message bubbles, issue report keyboard avoidance, voice haptics, and speech diagnostics

• Voice input has been upgraded with cloud speech recognition and a more stable recording, recognition, and text fill-in flow • Image editing now supports the same voice input workflow, making it easier to describe follow-up edits • User feedback reporting is more reliable, including queued report uploads when immediate sending is not available • Chat generation controls are more stable, with improved stop state handling and fewer accidental Return-key submissions during generation

• Image preview closing and memory handling are more stable, reducing leftover state when leaving previews • Generated image previews preserve their original aspect ratio more reliably • Images without file extensions are recognized more consistently and display correctly • Cached generated images can be sent back into editing more reliably, so continuing from an existing result is smoother

• The payment and recharge experience has been upgraded, with updated in-app purchase integration and smoother subscription and credits recharge flows • Generated image placeholders now adapt better to Dark Mode, including colors, borders, and progress feedback while images are being created • Remote attachment handling is more stable for image previews, file URLs, and file size information • Login and request authentication issues are easier to diagnose with clearer token and cookie diagnostics

• User image attachment previews now open faster by reusing local files and lighter preview data • Camera capture from the chat composer is more reliable, with clearer permission handling and fewer freezes • You can keep adding camera attachments while a chat response is being generated

• Tuned generated image preview sizing and cropping so image strips feel more natural across screen widths • Centralized invitation reward claim sheet presentation for more consistent open and close behavior • Reused shared conversation completion-state parsing to make synced conversation status more reliable • Restored the main app display name and refreshed invitation-related localization

• Generated image batches now appear as horizontal image strips, so finished images and in-progress placeholders stay together • Improved image generation status syncing for more timely preview updates when returning to a conversation or waiting for batch results • Chat sessions now reuse a shared WebSocket connection to reduce duplicate connections and state flicker during reconnects • Refined invitation reward pages and claim sheets in Dark Mode with clearer backgrounds and contrast

• Added a quick image editing action in chat so you can pick a photo from your library and start editing directly • Refined image preview editing controls and dismissal behavior for a smoother edit-from-preview flow • Improved the image preview drag-to-dismiss gesture to reduce accidental closes while browsing images • Made active reasoning state display more reliably while the model is thinking, with small chat UI refinements

• Added an edit action to generated image preview menus so you can continue refining an existing result • Added support for retrying a message in a new chat and improved attachment identity handling during retries • Improved conversation sync by reconciling remote completion state more reliably during recovery • Refined chat UI details, including sidebar hierarchy, main title bar layout, and quick action sheets

• Added an invite rewards flow so you can share an invite link and earn image credits when friends accept • Improved generated image result polling and message merging so completed images appear more reliably in chat • Stabilized prompt example loading by cancelling stale empty-state tasks when the chat view changes • Refreshed app icon/avatar artwork and restored the default translucent navigation bar style

• Upgraded generated image previews with a thumbnail strip plus save, share, and copy actions for easier multi-image browsing • Added an image edit prompt flow so you can continue refining generated images from the preview experience • Added a reasoning visibility setting to control whether reasoning content is shown in chat • Refined chat details, including clearer prompt gallery loading feedback and centered empty-state title text

• Added save-to-Photos and system share actions for generated image previews • Improved the prompt gallery with language-aware prompt selection and a larger iPad form sheet • Made connector action authorization more reliable by fetching the user profile before starting auth when needed • Refined chat UI details, including the template More button, sidebar divider, and dark/light appearance updates

• Added a built-in ecommerce image template carousel for quickly starting main image, scene, selling point, size, and other product image workflows • Added a More Inspiration entry that opens the prompt gallery and can start a new chat with the selected prompt • Improved generated image recovery when returning to the app, so async previews resume updating more reliably • Refined chat UI details, including a clearer selected conversation row and a simplified toolbar

• Generated images now appear as async previews directly in messages, making results easier to inspect as soon as they are ready • Updated tool localization and Tuist dependency support for clearer tool names and more reliable project maintenance