releases.shpreview
OpenAI/Release Notes

Release Notes

Mon
Wed
Fri
JulAugSepOctNovDecJanFebMarAprMayJunJul
Less
More
Releases43Avg13/mo

ChatGPT Work is a new agent for longer tasks that can research and analyze information, work across connected apps and files, and create finished documents, spreadsheets, presentations, reports, and Sites. Users can follow progress, ask questions, change direction, and approve actions. Work also supports Scheduled Tasks that run once, repeat on a schedule, trigger on events, or monitor for changes. The App Directory is being replaced with the Plugin Directory; existing app connections remain unaffected.

Read more →

The new ChatGPT desktop app is available globally for macOS and Windows, combining Chat for conversation, Work for research and deliverables with local file and app access, and Codex for software development with inline diff editing, pull-request review, faster computer use, and multi-repository support.

Read more →

Atlas is being deprecated as browser-based agentic capabilities are integrated into ChatGPT and Codex, with shutdown scheduled for August 9, 2026. Atlas browser data including bookmarks, open tabs, and browser history will not transfer automatically; users can export cookies and passwords to the ChatGPT desktop app and bookmarks to Chrome before the deadline.

Read more →

ChatGPT Sites lets you create dashboards, project trackers, launch calendars, prototypes, internal portals, and reports as interactive websites or lightweight apps, then preview and refine them with ChatGPT before sharing. Business and Enterprise customers can publish Sites publicly via URL; Pro, Pro Lite, and Edu users get beta access, with Plus users following in coming days. Free and Go plans are not supported, and public publishing is unavailable in the EEA, Switzerland, and the United Kingdom at launch.

Read more →

Beginning July 9, users on web, iOS, and Android will no longer be able to create new group chats, turn existing conversations into group chats, or join group chats through an invite link. Existing group chats remain available for now; when they become read-only, users retain access to shared messages, files, and images.

Read more →

ChatGPT Voice is now powered by GPT-Live-1 for paid users and GPT-Live-1 mini for Free users, both capable of listening and speaking at the same time for more natural turn-taking and interruptions. The model works inside ChatGPT chats with streamed text alongside spoken responses, supports web search and memory, and handles text and images in the same conversation. Rolling out across Free, Plus, and Pro plans on chatgpt.com and iOS/Android apps in supported regions; not available in Business, Enterprise, or Edu workspaces at launch, and does not support video or screen sharing.

Read more →

GPT-5.5 Instant Mini is now the fallback model for users hitting GPT-5.5 Instant or Auto rate limits, replacing GPT-5.3 Instant Mini. The model does not appear in the model picker and does not affect the API or Codex. Improvements include better tracking of evolving user intent, tone calibration, reduced repetitive or overly structured responses, stronger personalization, and fewer factual issues.

Read more →

Codex Remote is now generally available on all plans, with authenticated QR pairing for remote connections. The DigitalOcean Droplet Workspace plugin allows Codex to provision and connect a Droplet as a remote workspace. OpenAI o3 will be retired from ChatGPT on August 26, 2026, following a 90-day sunset period.

Read more →

Long pastes over 10,000 characters are now automatically converted to attachments in ChatGPT for Free and Go users, keeping the composer cleaner and preventing large pastes from consuming the context window. The threshold was also raised to 10k characters for Plus, Pro, and Business users, up from the previous 5k limit.

Read more →

As of June 12, 2026, GPT-5.2 models (Instant, Thinking, and Pro) are no longer available in ChatGPT. Existing conversations that used GPT-5.2 will automatically continue on the corresponding GPT-5.5 model.

Read more →

You can now delete memories from your memory summary page and turn memory off via the three-dot menu on web. If you do not want information from a chat used for personalization, you can turn off memory or use Temporary Chat. This is available to all ChatGPT users on web, and will be rolling out on mobile soon.

Read more →

Eligible Plus and Pro users gain rate-limit reset banking, and a developer mode is introduced for Chrome and in-app browsers, offering controlled DevTools Protocol access. New features also include an /init command for AGENTS.md scaffolding, customizable macOS Dock icons, an unread chats section, and clearer usage-limit errors.

Read more →

The model picker has been simplified to improve the balance of speed and reasoning effort, with new options including Instant, Medium, High, and plan-specific Pro choices. The picker now appears at the top of conversations on mobile and in the message composer on web. Users can now control auto-switching between Instant and Medium via Settings.

Read more →

ChatGPT can now turn some answers into interactive bar, line, pie, and scatter charts. Web users can enable table of contents in conversations longer than five responses, and writing blocks now support full-screen editing for essays, reports, and other documents with Library saving and download support. Fixed a long-standing issue where messages with attachments could not be edited.

Read more →

Lockdown Mode is now available as an opt-in advanced security setting across all account types and workspaces. When enabled, it restricts network-enabled capabilities including live web browsing, deep research, agent mode, file downloads, and some web-derived image support to reduce risk of data exfiltration from prompt injection attacks. Personal users enable it via Settings > Security, and workspace admins configure access through workspace settings and role-based access controls.

Read more →

ChatGPT's memory system now updates automatically to keep context current and reduce stale or contradictory saved memories, with users able to review memories through sources or memory summary. Plus and Pro users gain twice as much memory capacity. Rolling out to Plus and Pro users in the US today, expanding to Free and Go plans and additional countries over the next few weeks.

Read more →
Last Checked
1mo ago
Tracking since Dec 15, 2022