What's new
Latest Quad Chat updates: new AI model releases (GPT-6, Claude, Gemini, Grok), feature launches, performance improvements, and security updates.
2026-11-18: Connect your apps to Quad
- Open Connectors to see which apps are ready to connect.
- Choose read-only access or turn on write access with Auto approvals.
2026-11-18: Compare yearly and monthly plans
- Switch to monthly whenever you prefer a smaller upfront payment.
- See what your plan renews at before subscribing.
2026-11-17: Five new models in the chat picker
- Use GPT-6.2 Sol and Astra or Qwen4-Max on Pro and above.
- Try Inkling 2 Small and Muse Ember on any plan using regular chat credits.
2026-11-17: Faster startup on the web
- Open Settings or start with a collapsed sidebar without loading hidden chat history.
- See chat history as it arrives while projects and older groups finish loading.
- Reuse unchanged files when opening the app after an update.
2026-11-17: Six new ways to start a chat each day
- See ideas in your selected language, with built-in suggestions available when a daily set is still loading.
2026-11-17: A cleaner start on the Quad homepage
- See the product preview as soon as the homepage starts loading.
2026-11-17: More control over shared chats
- Review sharing details before creating a public link.
- Find, copy and remove shared links from Settings.
- Find chat details in a more compact panel and enjoy a refreshed sign-in page.
- Open a reply's sources beside the conversation, including available internal context.
- Change your appearance using a simpler, more compact menu.
- Choose from more themes with their own colors and type styles in light and dark mode.
- See your chosen accent in more places, including your messages and chat controls.
- Enjoy a refreshed mobile welcome screen with a gently animated background.
- Start a new mobile chat with a simpler screen.
- See your selected persona or recipe in a more compact composer control.
- Explore the mobile app with refreshed icons that fit your device.
- Choose more app themes and accent colors on mobile, synced with your account.
- Sign in on mobile with your email and password.
- Spot new features in the places you use them, starting with appearance themes.
- See Computer and browser activity in chat details as it happens, with the panel opening automatically until you close it.
- Keep conversations going more reliably when using free models.
- Use available free models on mobile even after your regular credits run out.
- See your daily free-model usage alongside your plan credits on mobile.
- Find free models directly from the mobile model menu.
- See clearer status messages when a briefing needs review.
- Scheduled reports manage their spending limits more reliably.
- Retry a credit-blocked reply after choosing a different model.
- Add files more reliably when using the mobile app in a browser.
- Review chat details and manage shared links from the mobile app.
- Open chat details from a compact menu with recognizable icons for web sources.
- See upcoming runs, recent activity and issues that need your attention in Automations.
- Run automations with GPT-6 Luna on every plan, including Free.
- Start a chat with more varied ideas for everyday tasks, learning and creative projects.
- Find stacks by the apps they include and browse skills with larger click targets.
- Read clearer translations across web and mobile.
2026-11-17: Your chat files in one place
- Show or hide chat details from the top bar, including on phones.
- Find files saved by Computer and open or download them beside your conversation.
- Review attachments, web sources and tasks scheduled in the chat. Load earlier items from longer conversations.
2026-11-16: Smoother replies in long chats
- Scrolling through long chats stays smoother while a reply is being written.
2026-11-16: Easier access to upgrades
- The desktop sidebar keeps your upgrade shortcut visible, including any offer your account qualifies for.
2026-11-16: More reliable replies
- Temporary provider failures get more chances to recover before the reply stops.
- Reconnecting to a reply continues from the current message instead of replaying earlier turns.
- Progress timing starts fresh for each reply, so a new request does not show time from an earlier answer.
- Beam can continue conversations whose earlier replies did not include reasoning.
- Provider refusals and unsupported requests show a clearer reason when a reply cannot continue.
2026-11-16: More reliable chat opening
- Context details show plain numbers when compact formatting is unavailable, so you can keep writing your message.
2026-11-15: Smoother page browsing
- Page artwork appears as you scroll and stays visible if animations cannot run.
2026-11-05: More reliable chat loading
- Chat context numbers use a safe language fallback if a language setting cannot be formatted.
- A failed page download keeps its original error while the page reloads, avoiding a second loading error.
2026-11-05: More reliable replies and page loading
- Fixed a saved-memory error that could stop background replies before they started.
- Safari can recover when a page file fails to load. Ordinary failed network requests keep their existing handling.
2026-11-04: Find the apps in your sidebar
- Open Download apps from the sidebar where Explore used to be. The extra download row above Upgrade is gone.
- Chats keep their history and tool results across the AI runtime update. Streamed replies no longer repeat the first fragment.
- Custom cron schedules run after a skipped daylight-saving time and continue through repeated hours. Daily schedules keep their existing behavior.
2026-11-04: The offer includes longtime free members
- The first subscription offer now includes older free accounts, with no seven-day signup cutoff.
- The upgrade page, sidebar and checkout use the same eligibility rules.
- The upgrade dialog shows the full three-month offer duration.
2026-11-04: Retry interrupted image downloads
- If an image download fails, you can try again from the image viewer. AI provenance marking stays attached to downloaded images.
2026-11-04: Clearer offers and working team settings
- Monthly pricing shows the new member discount, its duration and the regular price afterward.
- Close the offer banner to keep it hidden on this browser.
- Team settings now opens correctly instead of showing a page not found message.
2026-11-04: See the new member offer
- The new member offer appears above the workspace, plans page and public pages, with a link to view plans.
2026-11-04: New member offer and smoother loading
- Eligible new members can get 50% off their first three months on a monthly plan. The offer appears in the sidebar when it is available to you.
- The landing page shows a simpler loading screen before it is ready.
- More images can be added to chats when a browser does not identify the file type correctly.
2026-11-03: More reliable model choices
- Mercury 3 and other Standard Lane models correctly use your existing confirmation. Private chats keep zero-retention routing.
- Private models without an eligible zero-retention route stay out of the picker until one is available.
- PDFs use a compatible private model before your reply starts, and unavailable models show clearer recovery guidance.
- Failed replies stop cleanly without sending another cancellation after the provider has stopped.
2026-11-03: Smoother checkout
- Continue to checkout without the review dialog breaking when browser translation is on.
2026-11-01: More reliable phone connections
- Fixed a connection error that could keep an existing phone installation offline while its computer was online.
2026-11-01: More reliable policy confirmation
- Policy confirmation buttons keep working while saving, including when your browser translates the page.
2026-10-26: More reliable shared links and Claude replies
- Fixed a loading error affecting public Gallery links and creator profiles, including their link previews.
- Fixed a conversation format error that could stop Claude replies after Smart planning, including in existing conversations.
2026-10-26: New models for chats, coding and long documents
- Kimi K3.1 Pro, MiniMax M3.1 Flash, Mercury 3, Muse Spark Open and Nemotron 3.5 Super VL join the model picker.
- Older Kimi K3, MiniMax, Mercury and Meta Spark choices leave the picker. Existing chats keep their model history.
- Model prices, context limits and thinking controls reflect the latest available information.
2026-10-26: Public pages stay readable when loading is interrupted
- Public pages keep their text and links available when a page download fails, with readable headings in dark mode.
- Older translated home and features links now open the landing page in the same language.
2026-10-23: More reliable page loading
- Quad now recognizes more page-loading errors and can reload the app to recover during startup.
2026-10-22: Faster chats and clearer controls
- Long chats, earlier replies and open documents stay steady while a new answer streams. Workspaces, saved files and Studio collections load with fewer repeated requests.
- Web research reads independent pages together and avoids fetching or sending the same text twice. Other tools load when needed, with the same permissions.
- Smart chooses from GPT, Claude and Gemini within your plan and intelligence setting. Other available models can still be selected manually.
- For eligible difficult Smart requests, a premium model prepares a brief before Luna completes the work. Both models use credits at their own rates. If planning cannot finish, Smart continues with the original model. Manual choices and native audio or video requests keep their existing behavior.
- Open thinking to see Smart's task outline. Stats separates recorded charges from estimated savings, including when planning adds cost.
- Chats, temporary file analysis and Computer reuse supported prompt caches. Credit usage includes reported cache writes and partial responses, and combined response and summary charges round together.
- Read aloud stays beside Copy and Regenerate. Branch off, Rewrite, Check answer and Usage are together in the reply menu, and the newest answer keeps its actions visible.
- The context ring beside the model selector shows the last measured request, loaded skills and compaction details. Older document timestamps show days correctly, and automation usage labels follow your language setting.
- Drop files anywhere in a chat. The drop target clears when you leave or cancel, and dragging text keeps working normally.
- Opening Quad after an update recovers from earlier download failures. Public guides stay readable while the app loads.
- Updated guides, comparisons, Newsroom and support pages make product details easier to find. Starter ideas keep short labels and fill your composer with a detailed prompt when selected.
- Pricing and Upgrade explain monthly allowance renewals and estimated usage. AI reminders count active time and remember when you dismiss them.
- GitHub and Google Drive connections ask you to confirm in the same Quad account before saving access.
- Exported documents reject unsafe links while keeping their text and ordinary links available.
- GLM 5.3 FlashX is no longer offered in model pickers. GLM 5.3 Flash remains available, and Computer opens a Pro upgrade prompt when your plan does not include it.
2026-10-21: Privacy requests keep optional tracking paused
- The web app and updated mobile apps follow your account's privacy hold across devices.
- Cookie settings and email preference links cannot turn tracking back on while the hold is active.
- Resolving a request keeps tracking off until the hold is reviewed and you explicitly opt in again.
2026-10-21: More reliable privacy request emails
- Request receipts, data export links and deletion confirmations can reach accounts that store only a verified email address.
- Account deletion checks email contacts using that verified address too.
2026-10-18: Sign in with restricted browser storage
- Sign-in can continue when your browser blocks local storage. If authentication fails, you can try again.
- Your agreement is still required, and unavailable storage never counts as newsletter opt-in.
2026-10-17: Clearer recovery when a model is unavailable
- Unavailable Grok 4.1 Fast options are removed from new model choices. Existing chats keep their model details.
- When a provider reports that a model is unavailable, Quad asks you to choose another model instead of offering the same failed retry.
- Replies and usage already saved before a model failure are preserved.
2026-10-17: More reliable account review
- The account-review Continue button works with browser translation while saving your choices, including when you need to retry.
2026-10-16: More reliable sign-in
- The email sign-in button stays stable when browser translation changes its text and a loading indicator appears.
2026-10-15: More reliable PDFs and account checks
- Scanned PDFs and PDFs with a mix of scanned and text pages can be read in chat, up to eight pages at a time.
- If a chat request needs your 18+ confirmation or service permissions, the account check opens so you can review and complete it.
- Slow responses no longer leave you stuck with an eligibility error and no way to continue.
- Older mobile apps receive instructions for completing the account check in your browser.
2026-10-15: Clearer tables and more reliable chat
- Wide tables scroll sideways with readable columns in chat and fullscreen, including on phones.
- Dollar prices keep their formatting alongside equations, links and code.
- Citations update when sources arrive, and nested diagrams follow your theme.
- Large tool results keep their saved outcomes and sources instead of being cut in the middle.
- Temporary failures before a reply starts can retry safely. Saved unanswered messages have a Retry action.
- Expired sessions can refresh once, with a sign-in action if you need to log in again.
- Typing to focus the chat box handles more keyboard events without an error.
2026-10-14: Clearer plan details and better translations
- Pricing examples, team usage and upgrade messages reflect the current plan allowances. Web search shows credit-funded activity unless your team sets a custom cap.
- Support, account guidance, legal pages, chat controls and app previews use your chosen language more consistently.
- Longer translated buttons, tabs and menus fit more comfortably on small screens.
- Arabic and Hebrew navigation and translated public pages keep the correct reading direction.
2026-10-14: More relevant Knowledge search
- Project searches keep relevant passages from being crowded out by files in other projects.
- Knowledge search can rank passages by how well they answer your question.
- Background memory, list and check-in updates can skip exchanges that have nothing new to save.
- Check-ins are saved more reliably from longer conversations and messages in different languages.
2026-10-14: More reliable chat startup and image requests
- Chat can open in restricted browser frames that block offline features.
- Image requests in chat use the shapes supported by your selected image model. If a requested shape is unavailable, Quad asks you to choose before generating or charging credits.
2026-10-14: Interrupted chats keep their saved reply
- If a background chat is interrupted, Quad keeps the reply text it already saved when marking the turn as timed out.
- Long background chats keep their credit reservation active while the reply continues.
2026-10-14: More reliable Grok replies
- Chats using Grok 4.1 Fast no longer fail to start because Design or Computer command tools are available.
2026-10-14: Fresh searches for mobile suggestions
- Scheduled mobile suggestions can retrieve current web information instead of falling back when search could not start.
2026-10-14: More reliable memory and tool usage
- Quad is less likely to miss a memory update after your conversation.
- Some AI tool calls no longer count the same model cost twice.
2026-10-14: Guest chats work more reliably
- Messages sent from other networks no longer use up your guest chat allowance.
2026-10-14: Deleted chats no longer interrupt mobile
- Deleting a chat on another device no longer sends an older mobile app to an error screen.
2026-10-13: Quad checks in on what you mentioned
- A small line under the reply shows when Quad will check in, with Change and Don't.
- When it's time, Quad starts a chat itself and lets you know by notification, Telegram or the Briefing.
- Your open check-ins are listed in the Briefing, where you can mark them done, move them or remove them.
- Choose once a day, up to three a day, or off, and set quiet hours, in Settings under Notifications.
- Optionally let Quad prepare you for meetings from Google Calendar and tell you who's waiting on you in Gmail.
2026-10-13: Quad Agent and Quad Link for Mac and Linux
- Mac downloads work on Apple silicon and Intel. Linux downloads include AppImage and Debian packages for x64 and ARM64, requiring Ubuntu 22.04+, Debian 12+ or glibc 2.34+.
- The apps check for signed updates and let you restart to install them when your work is finished.
- Installed agents are detected automatically and appear in the model picker with their names.
- Choose from each agent's available models and supported reasoning options in the same picker.
- Agent conversations keep their native sessions when you reopen the app.
- Agent requires Quad sign-in for every connection, and signing out stops tasks and remote access.
- You can check whether your local server lists the model you selected before starting a task.
- Quad Link includes the same connection settings as Quad Agent.
2026-10-13: Fixes you might have hit, and safer agents
- Gemini 4 Argon is now featured in the model picker, so you can choose it without opening more families.
- Smart prefers Luna for coding and switches to Haiku when you ask it to fix a poor-looking interface.
- If a message fails to send, what you typed stays in the box.
- Quad can ask you a question with choices and continue when you answer.
- Scheduled tasks show the age, Terms and AI permission review link when needed, and return reserved credits before making an AI call.
- Image generation now starts with GPT Image 2.5 Flare, including on Free and Go.
- Updating the app in one tab no longer wipes a draft in another tab.
- Sharing a chat from the chat header works again.
- When a note, a report or a file save fails, you now see what happened and can retry.
- Note edits are saved even if you leave the note right away.
- Older clips in a saved storyboard play again.
- The Agent asks you before it submits a form or clicks something risky in the browser, and it can only open public web addresses.
- Turning off two-factor authentication now asks for a code from your authenticator app.
2026-10-13: More usage on Go and Pro, and clearer credits
- Go now includes 1,200 credits a month and Pro 2,000. That is more real usage than before, at the same price.
- Kimi K3 is available on Go and higher plans.
- A credit is now worth the same on every plan. A web search uses about 1 credit, and a typical Claude Sonnet message about 2.
- Web search no longer has a separate monthly allowance. It uses your credits like everything else, so it never runs out on its own.
- Your existing balance was converted to the new credits, so nothing was lost.
- When credits run low or run out, chat now says so right above the message box and shows when they come back.
- Free models keep working even when you are out of credits.
- A redesigned plans page, and an upgrade window you can open from anywhere.
- Pages open faster, and chats inside projects load faster.
2026-09-29: Docs, slides and designs
- Ask for a spec, a pitch deck or a one-pager and it opens beside the chat.
- Edit it together with Quad, then download it as Word, PowerPoint or PDF.
2026-08-25: Quad Chat for Android is on Google Play
- Download the native Android app free from Google Play
- Every chat, project and model synced with the web and iOS apps
- Cited web search, Image Studio and file attachments on mobile
- Voice input, push notifications and Android home-screen widgets
2026-07-31: Quad Chat for iPhone and iPad is on the App Store
- Download the native iOS app free from the App Store
- Every chat, project and model synced with the web app in real time
- Cited web search, Image Studio and file attachments on mobile
- Voice input, widgets and Live Activities for long-running answers
2026-06-03: A cleaner, faster workspace
- A cleaner design across chat, library, studio and settings
- Faster loads and smoother navigation throughout the app
2026-02-10: Skills
- Give the model short playbooks for documents, frontend design and writing
- Add your own instructions, or pull skills from a GitHub repo
2025-10-07: Faster pages
- Faster font loading
- Clearer page titles and descriptions in search results
2025-10-01: Official launch
- Access 75+ models including ChatGPT, Claude, Gemini, Grok, DeepSeek, and open models
- Persistent memory and web search included
- Image generation
- MCP integrations