LLM/AI Changelog

* **Gemini 4 Argon (`gemini-4-argon`) generally available (GA)**: Released
[Gemini 4 Argon](/docs/models/gemini-4-argon) (`gemini-4-argon`), our most
capable model built for complex coding, agentic, math, and science tasks.

For API changes, pricing, and migration guidance, see
[What's new in Gemini 4 Argon](/docs/whats-new-gemini-4-argon).

- **Gemini 4 Argon (`gemini-4-argon`) generally available (GA)** : Released
[Gemini 4 Argon](https://ai.google.dev/gemini-api/docs/models/gemini-4-argon) (`gemini-4-argon`),
our most capable model built for complex coding, agentic, math, and science
tasks.

For API changes, pricing, and migration guidance, see
[What's new in Gemini 4 Argon](https://ai.google.dev/gemini-api/docs/latest-model).

Composer predictions, available in beta, suggests your next message based on the conversation in your current Codex thread. After Codex responds, a suggestion may appear in the message box. Press Tab to accept it, then review or edit it before sending. Accepting a suggestion doesn’t send it automatically.
The beta is available to personal ChatGPT Pro users aged 18 and older in all supported regions, in the latest Codex desktop app. It supports local and SSH threads using GPT-6 Astra or GPT-6.1 Sol. Predictions are on by default; turn offShow predictionsin theComposersection ofGeneralsettings. During the beta, generating predictions doesn’t count toward Codex usage limits or consume credits. Messages you send, including accepted predictions, follow normal Codex usage limits and billing.
Read more:Composer predictions in Codex

* **Gemini 3.7 Flash deprecation**:
[`gemini-3.7-flash`](/docs/whats-new-gemini-3.7) is deprecated and has been
replaced by [`gemini-3.8-flash`](/docs/whats-new-gemini-3.8). It is no longer
available to select in Google AI Studio, and API requests to
`gemini-3.7-flash` are automatically routed to `gemini-3.8-flash`. Update
your model string to `gemini-3.8-flash` and review the
[Gemini 3.8 Flash guide](/docs/whats-new-gemini-3.8).

* **Gemini 3.5 Flash deprecation**:
[`gemini-3.5-flash`](/docs/whats-new-gemini-3.5) is deprecated and has been
replaced by [`gemini-3.6-flash`](/docs/whats-new-gemini-3.6). It is no
longer available to select in Google AI Studio, and API requests to
`gemini-3.5-flash` are automatically routed to `gemini-3.6-flash`. Update
your model string to `gemini-3.6-flash` (or
[`gemini-3.8-flash`](/docs/whats-new-gemini-3.8)) and review the
[Gemini 3.6 Flash guide](/docs/whats-new-gemini-3.6).

* **Deep Research Agent `deep-research-pro-preview-12-2025` deprecation**:
The `deep-research-pro-preview-12-2025` agent is deprecated and will be
[shut down](/docs/deprecations#managed-agents) on **October 23, 2026**.

Migrate your requests to one of the newer [Deep
Research](/docs/deep-research) agent versions:

* [`deep-research-preview-04-2026`](/docs/models/deep-research-preview-04-2026):
Designed for speed and efficiency, ideal to be streamed back to a client
UI.
* [`deep-research-max-preview-04-2026`](/docs/models/deep-research-max-preview-04-2026):
Maximum comprehensiveness for automated context gathering and synthesis.

To migrate, update the `agent` parameter in your `interactions.create`
requests from `deep-research-pro-preview-12-2025` to
`deep-research-preview-04-2026` (or `deep-research-max-preview-04-2026`).

- **Gemini 3.7 Flash deprecation** : [`gemini-3.7-flash`](https://ai.google.dev/gemini-api/docs/whats-new-gemini-3.7) is deprecated and has been replaced by [`gemini-3.8-flash`](https://ai.google.dev/gemini-api/docs/latest-model). It is no longer available to select in Google AI Studio, and API requests to `gemini-3.7-flash` are automatically routed to `gemini-3.8-flash`. Update your model string to `gemini-3.8-flash` and review the [Gemini 3.8 Flash guide](https://ai.google.dev/gemini-api/docs/latest-model).
- **Gemini 3.5 Flash deprecation** : [`gemini-3.5-flash`](https://ai.google.dev/gemini-api/docs/whats-new-gemini-3.5) is deprecated and has been replaced by [`gemini-3.6-flash`](https://ai.google.dev/gemini-api/docs/whats-new-gemini-3.6). It is no longer available to select in Google AI Studio, and API requests to `gemini-3.5-flash` are automatically routed to `gemini-3.6-flash`. Update your model string to `gemini-3.6-flash` (or [`gemini-3.8-flash`](https://ai.google.dev/gemini-api/docs/latest-model)) and review the [Gemini 3.6 Flash guide](https://ai.google.dev/gemini-api/docs/whats-new-gemini-3.6).
- **Deep Research Agent `deep-research-pro-preview-12-2025` deprecation** :
The `deep-research-pro-preview-12-2025` agent is deprecated and will be
[shut down](https://ai.google.dev/gemini-api/docs/deprecations#managed-agents) on **October 23,
2026**.

Migrate your requests to one of the newer [Deep
Research](https://ai.google.dev/gemini-api/docs/deep-research) agent versions:
- [`deep-research-preview-04-2026`](https://ai.google.dev/gemini-api/docs/models/deep-research-preview-04-2026): Designed for speed and efficiency, ideal to be streamed back to a client UI.
- [`deep-research-max-preview-04-2026`](https://ai.google.dev/gemini-api/docs/models/deep-research-max-preview-04-2026): Maximum comprehensiveness for automated context gathering and synthesis.

To migrate, update the `agent` parameter in your `interactions.create`
requests from `deep-re

- **Gemini 3.7 Flash deprecation** : [`gemini-3.7-flash`](https://ai.google.dev/gemini-api/docs/whats-new-gemini-3.7) is deprecated and has been replaced by [`gemini-3.8-flash`](https://ai.google.dev/gemini-api/docs/latest-model). All requests to `gemini-3.7-flash` are automatically routed to `gemini-3.8-flash`. Update your model string to `gemini-3.8-flash` and review the [Gemini 3.8 Flash guide](https://ai.google.dev/gemini-api/docs/latest-model).
- **Gemini 3.5 Flash deprecation** : [`gemini-3.5-flash`](https://ai.google.dev/gemini-api/docs/whats-new-gemini-3.5) is deprecated and has been replaced by [`gemini-3.6-flash`](https://ai.google.dev/gemini-api/docs/whats-new-gemini-3.6). All requests to `gemini-3.5-flash` are automatically routed to `gemini-3.6-flash`. Update your model string to `gemini-3.6-flash` (or [`gemini-3.8-flash`](https://ai.google.dev/gemini-api/docs/latest-model)) and review the [Gemini 3.6 Flash guide](https://ai.google.dev/gemini-api/docs/whats-new-gemini-3.6).
- **Deep Research Agent `deep-research-pro-preview-12-2025` deprecation** :
The `deep-research-pro-preview-12-2025` agent is deprecated and will be
[shut down](https://ai.google.dev/gemini-api/docs/deprecations#managed-agents) on **October 23,
2026**.

Migrate your requests to one of the newer [Deep
Research](https://ai.google.dev/gemini-api/docs/deep-research) agent versions:
- [`deep-research-preview-04-2026`](https://ai.google.dev/gemini-api/docs/models/deep-research-preview-04-2026): Designed for speed and efficiency, ideal to be streamed back to a client UI.
- [`deep-research-max-preview-04-2026`](https://ai.google.dev/gemini-api/docs/models/deep-research-max-preview-04-2026): Maximum comprehensiveness for automated context gathering and synthesis.

To migrate, update the `agent` parameter in your `interactions.create`
requests from `deep-research-pro-preview-12-2025` to
`deep-research-preview-04-2026` (or `deep-research-max-preview-04-2026`).

We’re rolling out faster steering in Codex in the ChatGPT desktop app. When you send a follow-up to steer a running task, Codex can respond to your changes sooner.
Use steering to correct an approach, add missing information, or change direction while Codex works. You can choose whether follow-up messages steer the current run or wait for the next one inSettings → General → Follow-up behavior.
Learn more inUsing Codex with your ChatGPT plan.

- **Gemini 3.7 Flash deprecation** : [`gemini-3.7-flash`](https://ai.google.dev/gemini-api/docs/whats-new-gemini-3.7) is deprecated and has been replaced by [`gemini-3.8-flash`](https://ai.google.dev/gemini-api/docs/whats-new-gemini-3.8). It is no longer available to select in Google AI Studio, and API requests to `gemini-3.7-flash` are automatically routed to `gemini-3.8-flash`. Update your model string to `gemini-3.8-flash` and review the [Gemini 3.8 Flash guide](https://ai.google.dev/gemini-api/docs/whats-new-gemini-3.8).
- **Gemini 3.5 Flash deprecation** : [`gemini-3.5-flash`](https://ai.google.dev/gemini-api/docs/whats-new-gemini-3.5) is deprecated and has been replaced by [`gemini-3.6-flash`](https://ai.google.dev/gemini-api/docs/whats-new-gemini-3.6). It is no longer available to select in Google AI Studio, and API requests to `gemini-3.5-flash` are automatically routed to `gemini-3.6-flash`. Update your model string to `gemini-3.6-flash` (or [`gemini-3.8-flash`](https://ai.google.dev/gemini-api/docs/whats-new-gemini-3.8)) and review the [Gemini 3.6 Flash guide](https://ai.google.dev/gemini-api/docs/whats-new-gemini-3.6).
- **Deep Research Agent `deep-research-pro-preview-12-2025` deprecation** :
The `deep-research-pro-preview-12-2025` agent is deprecated and will be
[shut down](https://ai.google.dev/gemini-api/docs/deprecations#managed-agents) on **October 23,
2026**.

Migrate your requests to one of the newer [Deep
Research](https://ai.google.dev/gemini-api/docs/deep-research) agent versions:
- [`deep-research-preview-04-2026`](https://ai.google.dev/gemini-api/docs/models/deep-research-preview-04-2026): Designed for speed and efficiency, ideal to be streamed back to a client UI.
- [`deep-research-max-preview-04-2026`](https://ai.google.dev/gemini-api/docs/models/deep-research-max-preview-04-2026): Maximum comprehensiveness for automated context gathering and synthesis.

To migrate, update the `agent` parameter in your `interactions.create`

Max and Team plans now include monthly API credits for running your own apps and agents on the Claude Platform. They roll out over a few days. Claim them by linking a Claude Console organization inSettings > Billing(Max) orOrganization settings > Billing(Team). Learn more aboutmonthly API credits for Max and Team plans.

* **Gemini Nano Banana 2.1 generally available (GA)**: Released
[Gemini Nano Banana 2.1](/docs/models/gemini-nano-banana-2.1)
(`gemini-nano-banana-2.1`), the latest high-efficiency image generation and
conversational editing model. An update to Nano Banana 2
([`gemini-3.1-flash-image`](/docs/models/gemini-3.1-flash-image)), it
maintains Flash-level speed and cost efficiency while delivering significant
improvements in visual quality, prompt adherence, multi-turn character
consistency, text rendering, and wide and panoramic aspect ratio generation
(`1:4`, `4:1`, `1:8`, `8:1`) across `1K`, `2K`, and `4K` resolutions.

See the [Image generation guide](/docs/image-generation) to get started.

* **Deprecation announcement**: The `gemini-3.1-flash-image` model is
deprecated (no shutdown date announced). Migrate to
[`gemini-nano-banana-2.1`](/docs/models/gemini-nano-banana-2.1).

- **Gemini Nano Banana 2.1 generally available (GA)** : Released
[Gemini Nano Banana 2.1](https://ai.google.dev/gemini-api/docs/models/gemini-nano-banana-2.1)
(`gemini-nano-banana-2.1`), the latest high-efficiency image generation and
conversational editing model. An update to Nano Banana 2
([`gemini-3.1-flash-image`](https://ai.google.dev/gemini-api/docs/models/gemini-3.1-flash-image)),
it maintains Flash-level speed and cost efficiency while delivering
significant improvements in visual quality, prompt adherence, multi-turn
character consistency, text rendering, and wide and panoramic aspect ratio
generation (`1:4`, `4:1`, `1:8`, `8:1`) across `1K`, `2K`, and `4K`
resolutions.

See the [Image generation guide](https://ai.google.dev/gemini-api/docs/image-generation) to get
started.
- **Deprecation announcement** : The `gemini-3.1-flash-image` model is
deprecated (no shutdown date announced). Migrate to
[`gemini-nano-banana-2.1`](https://ai.google.dev/gemini-api/docs/models/gemini-nano-banana-2.1).

Reference files saved in SharePoint libraries and OneDrive folders when creating a presentation with Copilot in PowerPoint
[Windows, Mac]
You can now reference files saved in SharePoint libraries and OneDrive folders when creating a presentation with Copilot in PowerPoint.
Roadmap ID:
555895
555897
Details:
What changed:
Copilot in PowerPoint can now recognize references to files stored in SharePoint and OneDrive when you mention them in your prompt, such as by file or document name. Previously, these implicit file references were not supported for files stored in these locations. This update makes it easier to use relevant Microsoft 365 content without manually attaching or uploading the file.
Why:
This change makes it easier to create presentations using content you already have in Microsoft 365. You can naturally reference a relevant file in your prompt, and Copilot can use that file as context when creating your presentation.
Try this:
Open PowerPoint and start creating a presentation with Copilot.
In your prompt, reference a file saved in SharePoint or OneDrive by the document name and/or owner of the document.
Ask Copilot to create a presentation using information from that file. For example: “Create a presentation summarizing the key findings from the DOCUMENT NAME.”
Why this matters:
You can bring existing SharePoint and OneDrive content into your presentation workflow more naturally, without first locating the file and manually attaching it.
Business impact:
Teams can create presentations faster by integrating cloud-stored resources without switching apps or manually uploading files.
Personal impact:
You save time locating and incorporating relevant files, reducing friction in presentation creation.

Seamless Search and Chat Integration
[Windows, Web]
Microsoft 365 Chat brings the power of conversational AI directly into Microsoft 365 Copilot Search, enabling users to move effortlessly from finding information to acting on it. Search results become the foundation for contextual chat, allowing you to ask follow-up questions, synthesize insights, and generate content—all in one unified experience. This integration reduces workflow fragmentation, accelerates decision-making, and delivers a more intuitive way to interact with your organization’s knowledge.
Roadmap ID:
512429
Details:
What changed:
Microsoft 365 Copilot Search now embeds conversational AI chat directly within search results. Previously, users had to switch between search and chat tools separately. This update allows follow-up questions, insight synthesis, and content generation based on search context in a unified interface, reducing workflow fragmentation.
Why:
Integrating chat with search streamlines user interaction with organizational knowledge, accelerating decision-making and reducing the need to switch tools.
Try this:
Open Microsoft 365 Copilot Search.
Enter a query and review the search results.
Use the chat pane to ask follow-up questions or generate content based on the results.
Why this matters:
This integration simplifies workflows by combining search and chat, helping users act on information faster.
Business impact:
Teams experience faster decision cycles and less disruption from switching between search and chat applications.
Personal impact:
You save time by accessing search results and conversational AI in one place, reducing task switching.

Regenerate response for alternative answers
[Web]
The Regenerate action lets you quickly get an alternative response to your latest prompt using options like Try Again or Switch Model, helping you explore different answers and improve output quality.
Details:
What changed:
Previously, users had to re-enter prompts to get different responses from Copilot. Now, the Regenerate action provides simple buttons to request alternative answers or switch AI models without starting over. This streamlines refining outputs and exploring variations.
Why:
This change reduces friction when iterating on AI-generated content, enabling faster exploration of ideas and improved response quality.
Try this:
Enter a prompt in Microsoft 365 Copilot Chat.
At the end of Copilot's response, open the "More options" menu.
Select 'Try Again' to get a new response.
Choose 'Switch Model' to see answers from a different AI model.
Why this matters:
Quickly generating alternative responses helps users find the best information or phrasing without repeating input steps.
Business impact:
Teams can iterate on AI-generated content more efficiently, improving decision-making and content creation speed.
Personal impact:
You save time by refining outputs with fewer steps and can explore more options easily.

Edit query string and user identity mapping rules for Copilot connectors
[Windows, Web]
Admins can now edit query string and user identity mapping rules directly in connector settings without opening a support ticket.
Details:
What changed:
Query string parameters and user identity mapping rules are now editable after a connection is created. Admins can update them in the Microsoft 365 admin center under Settings > Search & intelligence > Data sources, then select the connection and edit its configuration. Previously, these values were fixed at connection setup and changing them required support intervention.
Why:
Source systems change, teams re-scope projects, filters need tightening, and identity mappings drift as directory attributes are updated. Locking these rules at setup meant every adjustment became a support request, leaving connections stale or over-scoped in the interim.
Try this:
Go to the Microsoft 365 admin center > Settings > Search & intelligence > Data sources.
Select your connection (for example, Azure DevOps or ServiceNow) and open its configuration.
Update the query string or user identity mapping rule and save.
Confirm the change took effect on the next crawl and spot-check results for an affected user.
Why this matters:
Access scope and content scope are now self-serve. Admins can correct a mis-scoped connection or a broken identity mapping in minutes instead of waiting on a support cycle.
Business impact:
Faster remediation when a connection surfaces content to the wrong audience, fewer support tickets per connection, and configuration that keeps pace with changes in the source system—all of which support access-governance and compliance commitments.
Personal impact:
You fix scoping and permission-mapping issues yourself, on your own schedule, instead of filing a ticket and waiting for someone else to make a change.

Admin Center - detailed report on Everyone except external user permissions
[Web]
The new report provides SharePoint Advanced Management admins complete item-level details of permissions granted to special SharePoint groups: Everyone and Everyone except external users.
Roadmap ID:
561038
Details:
What changed:
The SharePoint Admin Center now includes a detailed report that shows item-level permissions for special SharePoint groups: Everyone and Everyone except external users. Previously, admins had limited visibility into these item-level permissions. This report helps identify exactly which items have these special SharePoint groups as recipients, improving permission auditing and security management.
Why:
Admins need precise visibility into permissions granted to broad groups to manage access risks effectively.

Enhance security compliance
Try this:
Open the SharePoint Admin Center and navigate to Advanced Management > All features > Data access governance reports.
Select the new report 'Sites and files shared via special SharePoint group' and trigger the relevant report.
Export the prepared report and review item-level permission details.
Why this matters:
Detailed permission reports help prevent unauthorized access and support compliance with organizational security policies.
Business impact:
Teams can better manage SharePoint permissions, reducing security risks and simplifying audits.
Personal impact:
Admins save time identifying and resolving permission issues with clearer, item-level insights.
Learn
Get item-level permission details for "Everyone except external users" and "Everyone"
Generate a report on sites and files shared via special SharePoint groups

Create custom skills in Copilot in PowerPoint
[Windows]
Create custom skills in Copilot in PowerPoint for tailored workflows for your specific presentation needs.
Details:
What changed:
Copilot now supports user defined custom skills, allowing users to extend its capabilities with personalized commands and workflows in PowerPoint. Previously, Copilot operated only with built-in skills and predefined functions in PowerPoint. This update enables customization within Microsoft 365 security and compliance standards.
Why:
Users need more flexibility to automate unique presentation tasks and workflows. Custom skills let users:

Enhance productivity by tailoring Copilot to their needs.
Try this:
Open PowerPoint and access the Copilot pane.
Select 'Create custom skill' to define new commands.
Test your custom skill within your presentation workflow.
Why this matters:
Custom skills let users personalize Copilot to match their unique presentation workflows, reducing manual effort.
Business impact:
Teams can standardize and automate specialized presentation tasks, improving consistency and efficiency.
Personal impact:
You save time by automating repetitive tasks and gain more control over Copilot’s assistance.

Reference files saved in SharePoint libraries and OneDrive folders when creating a presentation with Copilot in PowerPoint
[Mac]
You can now reference files saved in SharePoint libraries and OneDrive folders when creating a presentation with Copilot in PowerPoint.
Roadmap ID:
555897
Details:
What changed:
Copilot in PowerPoint can now recognize references to files stored in SharePoint and OneDrive when you mention them in your prompt, such as by file or document name. Previously, these implicit file references were not supported for files stored in these locations. This update makes it easier to use relevant Microsoft 365 content without manually attaching or uploading the file.
Why:
This change makes it easier to create presentations using content you already have in Microsoft 365. You can naturally reference a relevant file in your prompt, and Copilot can use that file as context when creating your presentation.
Try this:
Open PowerPoint and start creating a presentation with Copilot.
In your prompt, reference a file saved in SharePoint or OneDrive by the document name and/or owner of the document.
Ask Copilot to create a presentation using information from that file. For example: “Create a presentation summarizing the key findings from the DOCUMENT NAME.”
Why this matters:
You can bring existing SharePoint and OneDrive content into your presentation workflow more naturally, without first locating the file and manually attaching it.
Business impact:
Teams can create presentations faster by integrating cloud-stored resources without switching apps or manually uploading files.
Personal impact:
You save time locating and incorporating relevant files, reducing friction in presentation creation.

Copilot Analytics - Cowork adoption and impact
[Web]
Deeper insights into Cowork adoption and impact, accessible from Copilot dashboard and Consumption dashboard in the Microsoft 365 services tab. Includes export capability. All metrics can be queried by analysts with the new Consumption query (credit usage data) and through the Person query (Cowork usage data) in advanced insights.
Roadmap ID:
567005
Details:
What changed:
The Consumption dashboard and Copilot dashboard now include expanded metrics on Cowork adoption and its impact. Analysts can export data and use the new Consumption query in advanced insights to explore all metrics. Previously, such detailed analytics and export options were unavailable.
Why:
Providing deeper, exportable insights enables data-driven decisions about AI usage and investment, improving transparency and cost management.
Try this:
Open the Copilot dashboard in Copilot analytics.
Navigate to the Consumption dashboard > Microsoft 365 Services tab to view Cowork metrics.
Use the export option or run the Consumption query or Person query including Cowork usage metrics in advanced insights for detailed analysis.
Why this matters:
Better visibility into AI adoption helps organizations optimize usage and understand its business impact.
Business impact:
Teams gain actionable data to manage AI resource allocation and measure productivity improvements.
Personal impact:
You can analyze Cowork usage trends and costs more effectively, supporting informed decision-making.
Learn:
How to use the Consumption Dashboard in Insights

Upload audio files to ChatGPT to create transcripts, summarize recordings, and ask questions about their contents. You can turn a meeting, interview, or lecture into structured notes or a follow-up email draft. To get started, attach a supported audio file to a conversation and describe what you need.Audio uploads are available with paid ChatGPT subscriptions and workspaces. Availability may vary by workspace settings, region, client version, and model. Transcripts may contain errors, and performance may vary across languages.For more information, see:Uploading files and audio to ChatGPT.

Finances in ChatGPT is rolling out to Free and Go users in the U.S. on web, iOS, and Android. You can securely connect your financial accounts to find ways to save, understand your spending, and make sense of your investments, with answers grounded in your finances.
Learn more about Finances in ChatGPT.

It’s now easier to bring your notes and documents into ChatGPT on your phone. Capture multiple pages in a row, and ChatGPT automatically combines them into a single PDF, ready to upload to your chat.
To get started, open the camera in ChatGPT, tap the three-dot menu, and select Scan. Rolling out on iOS.

We’re introducing Pro 500, a new $500/month plan with the highest included usage of the Pro plans and access to Astra Ultrafast in ChatGPT Work and Codex. Select Ultrafast in the model picker to use a faster service tier for GPT-6 Astra. It uses your included allowance first, then your credit balance after that allowance is used.
Among Pro plans, Ultrafast is available only with Pro 500. Buying credits on Pro 100 or Pro 200 does not unlock it. Pro 500 subscriptions are available through ChatGPT on the web at launch.
Learn more aboutChatGPT Pro tiers.

Codex Cloud lets you start and continue coding tasks from desktop, web, or mobile using reusable environments with your project’s repositories, tools, and dependencies. Each task gets its own isolated workspace, and tasks can continue while your computer is asleep. In managed workspaces, your workspace’s Codex cloud-access settings apply.
To connect repositories and prepare an environment, see:Set up and configure Codex Cloud. For availability and usage, see:Using Codex with your ChatGPT plan.

Turn a conversation into a Page, start from a template, or begin writing directly in Space. Edit the text yourself or ask ChatGPT to revise it, add charts, or create interactive content. Collaborators can work on the same Page and leave comments, with each person using their own ChatGPT.
Pages follow Space’s availability and sharing settings. Sharing a Page does not give other people access to your private chats or memory; information added to the Page is visible to people who can view it.

Site owners and editors on Plus and Pro can selectEdit sitefrom a published Site in a desktop browser to continue editing in ChatGPT, review changes, and publish when ready.
Site owners can also create recurring cloud schedules fromAutomations. These updates are available where Sites is supported. Learn more aboutcreating and managing Sites.

We’re introducing dots: always-on agents in ChatGPT that can take on ongoing work and keep making progress between conversations. Give your dot a goal, connect the apps it needs, and define what it can do on its own. Powered by GPT-6 Astra, your dot has its own cloud computer and can bring results back for your review.
Dots are rolling out gradually to eligible Pro and Business Premium users aged 18 and older in supported markets. Pro access excludes the European Economic Area, Switzerland, and the UK at launch. Enterprise access is a beta and is off by default. Create your dot in the ChatGPT desktop app or on desktop web.
For the next month, dots usage won’t count toward eligible Pro, Business, and Enterprise users’ plan allowances. After this period, we’ll share usage terms for each plan.
Learn more aboutgetting started with your dot.

We’re making it easier to bring meeting context into ChatGPT so you can act on next steps faster. The Meetings plugin takes notes for you based on what’s said in the meeting and saves personalized summaries with action items in ChatGPT Space. You can keep them private or share them with your team, then ask ChatGPT to update a project plan or draft a follow-up.
Audio is deleted once your notes are ready and can’t be accessed or replayed. Available in beta in the ChatGPT desktop app on macOS to Pro and Business users, with Enterprise coming soon.Learn more.

Space brings your Pages, files, and related work together in ChatGPT. Create a space for a project, add the material you need, and share it with other people so you can keep building on the same work. Space replaces Library for accounts with access; your existing Projects remain separate.
Available to eligible Pro, Business, and Enterprise users in the ChatGPT desktop app and on the web. Sharing in managed workspaces follows your organization’s settings; Enterprise sharing is an opt-in preview.

Use Code Review in Codex to read pull-request summaries and diffs, review findings, and discuss changes. Open a pull request, inspect the changes, and ask Codex to explain a section or help address a potential issue before sharing feedback.
Available through the Code Review plugin on supported desktop and web experiences. Access to a repository follows your connected account and workspace permissions.
Learn more aboutCode Review.

Codex Security Cloud can scan GitHub repositories on demand or on a schedule, while ongoing checks review new commits. Codex investigates findings, removes duplicates, and prepares proposed fixes for review. Scans run in the cloud and can continue after you close your laptop.
Access requires the applicable workspace permissions, repository setup, and billing configuration. Existing Codex Security customers opt in before paid usage. This feature provides access to Daybreak Blue models only within Codex Security Cloud.
Learn more aboutsetting up Codex Security Cloud.

You can use your ChatGPT account to sign in to participating products without creating another password. The product receives your name, email address, and profile picture if you have one. Signing in does not share your ChatGPT conversations or memory; you approve any additional access separately.
Available to signed-in ChatGPT users, subject to organization settings. Sign-in access is separate from permission to use your ChatGPT plan allowance.
Learn more aboutSign in with ChatGPT.

Supported plugins can use MCP events to start an automation when something changes in a connected app. Describe what you want ChatGPT to follow and what it should do with matching updates—for example, prepare a summary when new information arrives in a channel you have asked it to monitor.
The connected account must have access to the event and any information used by the task.
Learn more aboutconnected apps in ChatGPT.

Eligible Plus and Pro users can authorize participating apps to use their ChatGPT model allowance without creating an API key. Choose which apps can use your allowance and set an app-specific limit. Usage counts toward your existing plan allowance; connecting an app does not add more usage.
This is a limited preview with participating integrations. Using purchased credits requires a separate opt-in. A partner may still charge for its own product or services.
Learn more aboutusing your ChatGPT allowance in other apps.

Plugins can now include interactive panels, sidebar views, and viewers or editors for supported file types. Work with an app alongside your conversation, select the information you want ChatGPT to use, or complete a form provided by the plugin.
Plugins still require their applicable account connections and workspace permissions.
Learn more aboutplugins in ChatGPT and Codex.

We’re introducing security history, a new way to review recent security activity for your OpenAI account. You can see sign-ins, sign-outs, and changes to multi-factor authentication (MFA), passkeys, and other security settings. Events include the time, location, and device details. Some details may be approximate or unavailable.
To view your history in ChatGPT on the web, go toSettings > Security and login, then selectSecurity history. For help reviewing unfamiliar activity and securing your account, see:Keeping your OpenAI account secure.

What
: Added the
creator
property (noting support for
author
) and updated
interactionStatistic
to document supported interaction types in the
VideoObject structured data documentation
.
Why
: To document support for the
creator
and
author
properties, and clarify supported interaction types for
interactionStatistic
in VideoObject structured data.

Navigate to your workbook changes directly from the Copilot chat pane response
[Windows, Web, Mac]
In Copilot in Excel, the response contains links to highlight where changes are made in your workbook. Navigate directly to new sheets, tables, ranges, and objects like charts and shapes.
Details:
What changed:
In the Copilot chat response, there now is a link you can click that will take you directly to changes in the workbook. Previously, you would need to locate the change in the response and navigate in the sheet on your own.
Why:
We want you to understand where copilot made edits and connect you to them directly from the chat response. This will help build confidence in the changes and allow you to review while you're reading the response in the chat pane.
Try this:
Send a prompt to Copilot where the workbook is modified.
The chat response will show the changes in a different format with a highlight to indicate a link you can click for navigation. New Sheets, tables, ranges, and objects will be highlighted.
Why this matters:
The feature allows quick review of document changes made by Copilot, increasing transparency of the edits.
Business impact:
Teams can understand changes made while reading the response from Copilot and quickly navigate to those places for review.
Personal impact:
You can review the Copilot Edits in your document.