OpenAI
npx @buildinternet/releases get openaiChatGPT Ads now live in 31 European markets
This release5 featuresNew capabilitiesAI-tallied from the release notesChatGPT Ads is expanding to 31 European countries, including Germany, France, Spain, Italy, and the Netherlands, following a six-month U.S. pilot. Advertisers start through the sales team or partners, with self-service Ads Manager arriving later this summer. Ads appear only on Free and Go plans; Plus, Pro, and Enterprise stay ad-free. The platform also adds conversion optimization, geo-targeting, custom audiences, and expanded measurement.
Named data-residency endpoints supported
This release1 featureNew capabilities1 fixBug fixesAI-tallied from the release notesAdded support for named data-residency endpoints. Also requires patched optional networking dependencies to address a bug fix.
ChatGPT for Teens ships with Study Mode, parent controls
This release6 featuresNew capabilitiesAI-tallied from the release notesOpenAI introduces ChatGPT for Teens, an experience for users 13–17 with age-appropriate safeguards on by default, including Study Mode, responsible homework reminders, quizzes, Learning Visualizations, and Study Hours. Adds new Parental Controls and a partnership with CodeAI to bring AI education to students and teachers.
MCP forms and editable approvals land; voice and diff review fixed
This release5 featuresNew capabilities4 enhancementsImprovements to existing features6 fixesBug fixesAI-tallied from the release notesAdded a setting to open ChatGPT directly in Codex Remote on launch, plus support for standard MCP forms and editable message approvals. Fixed large task responses failing to load, tasks disappearing after idle or reconnect, and improved diff review stability and voice reliability.
Bedrock Runtime endpoints supported
Breaking (minor)This release1 featureNew capabilitiesAI-tallied from the release notesAdds endpoint support for Bedrock Runtime, expanding the SDK's integration surface.
7.5.0 (2026-08-17)
Bedrock Runtime support; shell call streaming events added
This release2 featuresNew capabilitiesAI-tallied from the release notesAdded Bedrock Runtime endpoint support and shell call streaming events with new service/image types.
Adds opt-in gateway /v1/models discovery for the model picker, a --plugin-url flag for loading plugins from archives, new configurable environment variables for synchronous plugin updates and package-manager override, and MCP server plugin callbacks for tool responses. Also improves startup clarity by warning on unknown config keys and tightening file uploads to require an explicit content type.
WebSocket stream IDs; Sora video APIs deprecated
Breaking (minor)This release4 featuresNew capabilities1 enhancementImprovements to existing featuresAI-tallied from the release notesAdds WebSocket stream IDs and an Ultrafast tier, plus structured MCP and websocket errors. Sora video APIs are now deprecated.
Project memory settings editable after creation
This release1 featureNew capabilitiesAI-tallied from the release notesYou can now change a project's memory setting after creation, choosing between Default memory and Project-only memory. Project-only memory keeps conversations and context scoped within the project, and is the default for shared projects (which cannot switch).
Computer History for macOS tracks app activity to resume context
This release1 featureNew capabilitiesAI-tallied from the release notesAn optional Computer History feature in the ChatGPT macOS app lets ChatGPT and Codex reference selected app and website activity, helping users continue work without re-explaining context. Available to Pro, Business, and Enterprise users.
App directory ships; connectors now listed as apps
This release1 featureNew capabilities1 enhancementImprovements to existing featuresAI-tallied from the release notesYou can now browse and add approved apps in the new app directory, which lets you work with tools and data directly inside conversations. Connectors now appear in the directory as apps, consolidating all tool management in one place.
Ultrafast tier runs GPT-5.6 Sol 14× faster
This release1 featureNew capabilitiesAI-tallied from the release notesOpenAI previews Ultrafast, a new service tier running GPT-5.6 Sol up to 14× faster than Standard processing, with up to 750 output tokens per second. Launching first in the OpenAI API, Ultrafast is powered by Cerebras and targets time-sensitive workflows like incident response and reliability.
Enterprise and Edu users will have 10 deep research queries per month. Look for the tool name in the response: {"tool_name":"deep_research"}.
ChatGPT for Intune is a separate iOS app for ChatGPT Enterprise organizations that manage mobile access management through Microsoft Intune and Microsoft Entra. The app lets IT teams apply Microsoft app protection policies and Conditional Access policies to the ChatGPT mobile…
Updating projects with a new tabbed layout so users can easily access sources alongside their chats.
Company knowledge will support custom MCP connectors with search/fetch functionality. Custom connectors fitting these criteria that are enabled for your organization will appear in the company knowledge experience for allowed users, giving teams more complete, accurate, and…