LLM Changelog

Admins and users can now build and deploy custom connectors to proprietary systems using Model Context Protocol (MCP).
Requires a remote MCP serverAvailable only in deep researchAdmin-published connectors appear in the connector list for all users
Learn more aboutbuilding custom connectors with MCP. For Team, Enterprise, and Edu plans, only admins can build and deploy custom connectors.

Memory improvements are starting to roll out to Free users. In addition to the saved memories that were there before, ChatGPT now references your recent conversations to deliver responses that feel more relevant and tailored to you.
Free users must be logged in and on up-to-date apps (iOS/Android v1.2025.147+).
Opt‑in reminders
Free users in EEA (EU + UK), Switzerland, Norway, Iceland, or Liechtenstein will see a prompt to enable this setting or can visitSettings>Personalization>Memory>Reference chat historyto enable.Outside the European regions listed above, all Free users that have memory enabled will receive the upgrade automatically.
You can turn off memory anytime in settings. Learn more in ourMemory FAQ.

Insert images in adaptive cards for richer interactions
[Windows, Web]
Make your adaptive cards more dynamic by adding images—perfect for illustrating ideas, sharing visual data, or engaging users with eye-catching content. This feature helps teams communicate clearly, support diverse learning styles, and create more memorable interactions in everyday workflows.
Learn more
.

Reuse connector and API actions across multiple copilots
[Web]
Take an action that already works in Copilot for Sales and publish it to Customer Service—or any other eligible Copilot—in a few clicks. Skip duplicate setup, speed up delivery, and keep functionality consistent, all with built-in admin approval flows.
Learn more
.

Unpublish connector actions when data or services change
[Web]
Roll back a published connector to draft so it no longer appears in Microsoft 365 Copilot. Makers can pause their own connectors, and admins can disable any connector to update configurations or retire obsolete services—without deleting them.
Learn more
.

Copilot in Forms can now reference files to help generate drafts
[Web]
When creating a form with Copilot, users can now reference existing documents such as Word, Excel, and PowerPoint. Users can also reference an existing form by pasting the form's URL into the prompt box. Additionally, Copilot can search and suggest relevant files when generating a form, so users can easily create a draft that meets their needs.
Learn more
.

Copilot in Forms has been refreshed to help users better refine and edit their forms
[Web]
Copilot in Forms has been revamped to more easily help users refine and modify their forms. Users can now type prompts to Copilot to help with editing and refinement, so they can get tailored suggestions and easily get their forms ready to send.
Learn more
.

Usage reports – monitor m365.cloud.microsoft/chat/Teams/Outlook activity in Microsoft 365 Copilot Chat usage report
[Web]
Track active users and last activity dates for m365.cloud.microsoft/chat, Teams, and Outlook—even for employees without a Copilot license—to gauge grassroots adoption and fine-tune rollout plans in Microsoft 365 Copilot Chat usage report.
Learn more
.

Analyze usage of declarative agent in Developer Portal
Developers of custom app and published apps can analyze declarative agent usage in Developer Portal. By using these analytics, developers can gain valuable insights into how users interact with your app, identify areas for improvement, and make data-driven decisions to enhance the overall user experience.
Learn more
.

Insert images in adaptive cards for richer interactions
[Windows, Web]
Make your adaptive cards more dynamic by adding images-perfect for illustrating ideas, sharing visual data, or engaging users with eye-catching content. This feature helps teams communicate clearly, support diverse learning styles, and create more memorable interactions in everyday workflows.
Learn more
.

Reuse connector and API actions across multiple copilots
[Web]
Take an action that already works in Copilot for Sales and publish it to Customer Service-or any other eligible Copilot-in a few clicks. Skip duplicate setup, speed up delivery, and keep functionality consistent, all with built-in admin approval flows.
Learn more
.

Unpublish connector actions when data or services change
[Web]
Roll back a published connector to draft so it no longer appears in Microsoft 365 Copilot. Makers can pause their own connectors, and admins can disable any connector to update configurations or retire obsolete services-without deleting them.
Learn more
.

Usage reports - monitor m365.cloud.microsoft/chat/Teams/Outlook activity in Microsoft 365 Copilot Chat usage report
[Web]
Track active users and last activity dates for m365.cloud.microsoft/chat, Teams, and Outlook-even for employees without a Copilot license-to gauge grassroots adoption and fine-tune rollout plans in Microsoft 365 Copilot Chat usage report.
Learn more
.

**API updates:**

- Launched support for [custom video preprocessing](https://ai.google.dev/gemini-api/docs/video-understanding#customize-video-processing) using clipping intervals and configurable frame rate sampling.
- Launched multi-tool use, which supports configuring [code execution](https://ai.google.dev/gemini-api/docs/code-execution) and [Grounding with Google Search](https://ai.google.dev/gemini-api/docs/grounding) on the same `generateContent` request.
- Launched support for [asynchronous function calls](https://ai.google.dev/gemini-api/docs/live-tools#async-function-calling) in the Live API.
- Launched an experimental [URL context tool](https://ai.google.dev/gemini-api/docs/url-context) for providing URLs as additional context to prompts.

**Model updates:**

- Released `gemini-2.5-flash-preview-05-20`, a Gemini [preview](https://ai.google.dev/gemini-api/docs/models#model-versions) model optimized for price-performance and adaptive thinking. To learn more, see [Gemini 2.5 Flash Preview](https://ai.google.dev/gemini-api/docs/models#gemini-2.5-flash-preview) and [Thinking](https://ai.google.dev/gemini-api/docs/thinking).
- Released the [`gemini-2.5-pro-preview-tts`](https://ai.google.dev/gemini-api/docs/models#gemini-2.5-pro-preview-tts) and [`gemini-2.5-flash-preview-tts`](https://ai.google.dev/gemini-api/docs/models#gemini-2.5-flash-preview-tts) models, which are capable of [generating speech](https://ai.google.dev/gemini-api/docs/speech-generation) with one or two speakers.
- Released the `lyria-realtime-exp` model, which [generates music](https://ai.google.dev/gemini-api/docs/music-generation) in real time.
- Released `gemini-2.5-flash-preview-native-audio-dialog` and `gemini-2.5-flash-exp-native-audio-thinking-dialog`, new Gemini models for the Live API with native audio output capabilities. To learn more, see the [Live API guide](https://ai.google.dev/gemini-api/docs/live-guide#native-audio-output) and [Gemini 2.5 Flash Native Audio](https://ai.google

Create custom practice quizzes with GeminiWhat:Starting today, you can generate custom practice quizzes to help you prepare for an upcoming exam or simply increase your knowledge of any topic - big or small. Create quizzes based on documents you want to study, such as PDFs or class notes, or ask Gemini to create a quiz on a specific topic, and you’ll get a dynamic quiz experience, complete with hints, explanations for right and wrong answers, and a helpful summary at the end highlighting where you did well and where you may need to study a little harder. This experience is available to users over the age of 18 and to qualifying Google Workspaceeducationplans.Why:With Gemini’s intuitive and streamlined practice quizzes, learners can lean on the advantages of generative AI to help them prepare: unlimited quiz generation, personalized responses, and conversational learning experiences.Create with CanvasWhat:Discover new ways to create in Canvas! Starting today, the new Create menu lets you transform text into a variety of dynamic content, custom web pages, visual infographics, engaging quizzes, and immersive Audio Overviews. Or describe anything that you want to create and watch Gemini generate code to build a working prototype. Then collaborate with Gemini to customize it to your needs.Vibe coding apps in Canvas just got better too! With just a few prompts, you can now build fully functional personalized apps in Canvas that can use Gemini-powered features, save data between sessions, and share data between multiple users. You can even save a shortcut to your apps on your phone home screen for easy access. Lastly if there are errors in the app, Canvas will automatically try to resolve them for you.Why:Whether you’re writing, visualizing information, or vibe coding personal apps, Canvas helps you transform a blank slate into a share-worthy creation in minutes. Focus on your vision to create something awesome and leave the heavy lifting of generating, editing, and fixing

Create custom practice quizzes with GeminiWhat:Starting today, you can generate custom practice quizzes to help you prepare for an upcoming exam or simply increase your knowledge of any topic – big or small. Create quizzes based on documents that you want to study, such as PDFs or class notes, or ask Gemini to create a quiz on a specific topic, and you'll get a dynamic quiz experience, complete with hints, explanations for right and wrong answers and a helpful summary at the end highlighting where you did well and where you may need to study a little harder. This experience is available to users over the age of 18 and to qualifying Google Workspaceeducationplans.Why:With Gemini’s intuitive and streamlined practice quizzes, learners can lean on the advantages of generative AI to help them prepare: unlimited quiz generation, personalized responses, and conversational learning experiences.Create with CanvasWhat:Discover new ways to create in Canvas! Starting today, the new Create menu lets you transform text into a variety of dynamic content, custom web pages, visual infographics, engaging quizzes and immersive Audio Overviews. Or describe anything that you want to create and watch Gemini generate code to build a working prototype. Then collaborate with Gemini to customise it to your needs.Vibe coding apps in Canvas just got better too! With just a few prompts, you can now build fully functional personalised apps in Canvas that can use Gemini-powered features, save data between sessions and share data between multiple users. You can even save a shortcut to your apps on your phone home screen for easy access. Lastly, if there are errors in the app, Canvas will automatically try to resolve them for you.Why:Whether you're writing, visualising information or vibe coding personal apps, Canvas helps you transform a blank slate into a share-worthy creation in minutes. Focus on your vision to create something awesome and leave the heavy lifting of generating, editing, and fixi

Since its launch in the API in April, GPT-4.1 has become a favorite among developers—by popular demand, we’re making it available directly in ChatGPT.
GPT-4.1 is a specialized model that excels at coding tasks. Compared to GPT-4o, it's even stronger at precise instruction following and web development tasks, and offers an alternative to OpenAI o3 and OpenAI o4-mini for simpler, everyday coding needs.
Starting today, Plus, Pro, and Team users can access GPT-4.1 via the "more models" dropdown in the model picker. Enterprise and Edu users will get access in the coming weeks. GPT-4.1 has the same rate limits as GPT-4o for paid users.

GPT-4.1 mini is a fast, capable, and efficient small model, delivering significant improvements compared to GPT-4o mini—in instruction-following, coding, and overall intelligence. Starting today, GPT-4.1 mini replaces GPT-4o mini in the model picker under "more models" for paid users, and will serve as the fallback model for free users once they reach their GPT-4o usage limits. Rate limits remain the same.
Evals for GPT-4.1 and GPT-4.1 mini were originally shared in theblog postaccompanying their API release. They also went through standard safety evaluations. Detailed results are available in the newly launchedSafety Evaluations Hub.

Single connector for knowledge and actions
[Web]
You can now reuse connector actions across multiple Copilot deployments without recreating them each time. This feature lets you select a previously published action—such as one from Copilot for Sales—and publish it to other endpoints, such as Copilot for Customer Service, with just a few clicks. It's enabled by default and streamlines deployment while reducing duplication.
Learn more
.