LLM/AI Changelog

- Released [Veo 3.1 and 3.1 Fast](https://ai.google.dev/gemini-api/docs/video#veo-3.1) models in
public preview, with new features including:

- Extending Veo-created videos.
- Referencing up to three images to generate a video.
- Providing first and last frame images to generate videos from.

This launch also added more options for Veo 3 output video durations: 4, 6,
and 8 seconds.
- Deprecation announcement: Shut down for `veo-3.0-generate-preview` and
`veo-3.0-fast-generate-preview` coming November 12, 2025.

ChatGPT now manages saved memories automatically by keeping the most relevant details prioritized and moving less important ones to the background. This helps prevent saved memories from reaching capacity and helps avoid “memory full” state in your ChatGPT account.
To decide which memories stay top of mind, ChatGPT considers factors such as how recent a detail is and how often you talk about a topic. You can now also search your saved memories more easily and sort them by newest or oldest.
You’re always in control: you can turn off automatic memory management at any time, and you can see which memories are currently top of mind and choose to prioritize or deprioritize any specific memory yourself in settings. You can also view and restore prior versions of saved memories.
These updates are designed to make saved memory feel seamless, reliable, and easy to manage. We’re beginning to roll this out to Plus and Pro users globally on the web.

Powered by Haiku 4.5:Claude in Chrome now defaults to Haiku 4.5 so it’s a faster, more responsive experience. You can always switch back to Sonnet 4.5.Claude handles image uploads for you:Give Claude an image and tell it where to upload, whether it’s an expense report, form attachment, or a picture upload.Show Claude exactly what you mean:Take a screenshot or drag to highlight specific parts of your screen. Point Claude to the exact button, field, or detail—much faster than describing complex layouts in words.

Historical Data Upload Support in Organizational Data in Microsoft 365
[Web]
Enhance your data management by uploading historical HR data manually via CSV files. If the historical option is selected, admins can assign an effective date for precise processing by apps like Viva Insights. This ensures consistent and accurate data across Microsoft 365 and Viva apps.
Learn more.

We're excited to announce that ChatGPT Go, our low cost subscription plan, is now available in71 additional countries, bringing total coverage to89 countries.
For an affordable price, ChatGPT Go provides everything included in theFreeplan, and:
More messages, larger file uploads, and expanded image generationAccess to advanced data analysisLonger memory for more personalized responses
Available on web, mobile (iOS & Android), and desktop (macOS & Windows).
Note:In Cambodia, Laos, and Nepal, ChatGPT Go subscriptions aren’t available in the iOS app currently. Web and Android subscriptions remain available.
Learn more aboutChatGPT Go.

We’re rolling out two new ways to bring Slack and ChatGPT together: aconnectorthat brings your Slack context into ChatGPT, and a ChatGPT app for Slack – an integration that enables you to chat with ChatGPT from inside Slack.
With theChatGPT app for Slack,you can chat one-on-one with ChatGPT in a dedicated Slack sidebar – a space to ask questions, summarize long threads into action items, draft replies, and search messages and files you already have access to. Chats you start in Slack also appear in your ChatGPT sidebar, so it’s easy to pick up later from web or mobile. Semantic search is supported for Slack customers with AI enabled on Business+ or Enterprise+ plans; all other plans use keyword search.
With theChatGPT connector for Slack,you can securely bring in context from your Slack channels and DMs, making responses more helpful Available in chat, Deep Research, and Agent Mode. Users can enable fromSettings > Apps & Connectors.
Both app and connector are available toPlus, Pro, Business and Enterprise/Educustomers. Additionally, the ChatGPT app for Slack requires apaid Slack account;availability and workspace installation may depend on your Slack workspace settings.
Installing the app requires the connector to be enabled for your account.
Visit theChatGPT app for Slackpage to get started with installation.

We're addingNotionandLinearas synced connectors (for Pro users), so you can securely bring Notion pages and Linear issues/discussions into Chat for fast answers and summaries. You can enable sync for Notion and Linear from your Connector settings (connect to Notion or Linear if you haven’t already, first). Once synced, ChatGPT will automatically reference the indexed content when relevant. Learn more aboutsynced connectors.

We're excited to announce that ChatGPT Go, our low cost subscription plan, is now available in16 additional countries. For an affordable price, ChatGPT Go provides everything included in theFreeplan, and:
More messages, larger file uploads, and expanded image generationAccess to advanced data analysisLonger memory for more personalized responses
Available on web, mobile (iOS & Android), and desktop (macOS & Windows).
Note:In Cambodia, Laos, and Nepal, ChatGPT Go subscriptions aren’t available in the iOS app currently. Web and Android subscriptions remain available.
Learn more aboutChatGPT Go.

Updating GPT-5
We’re updatingGPT-5 Instantto better recognize and support people in moments of distress.
The model is trained to more accurately detect and respond to potential signs of mental and emotional distress. These updates were guided by mental health experts, and help ChatGPT de-escalate conversations and point people to real-world crisis resources when appropriate, while still using language that feels supportive and grounding.
As we shared in arecent blog, we've been using our real-time router to direct sensitive parts of conversations—such as those showing signs of acute distress—to reasoning models. GPT-5 Instant now performs just as well as GPT-5 Thinking on these types of questions. When GPT-5 Auto or a non-reasoning model is selected, we'll instead route these conversations to GPT-5 Instant to more quickly provide helpful and beneficial responses. ChatGPT will continue to tell users which model is active when asked.
This update to GPT-5 Instant is starting to roll out to ChatGPT users today.We’re continuing to work on improvements and will keep updating the model to make it smarter and safer over time.

We’re rolling out parental controls globally in ChatGPT to help families guide how ChatGPT works in their homes. Available starting on the web today, mobile soon.
To set up parental controls, a parent or guardian sends an invite to their teen to connect accounts. After the teen accepts, the parent can manage the teen’s settings from their own account. Teens can also invite a parent to connect.
Once a teen is connected to a parent’s account, they automatically receive additional content protections, such as reduced graphic content and viral challenges. Parents will have the option to turn this setting off if they choose, but teen users cannot make changes.
Parents don’t have access to their teen’s conversations, except in rare cases where our system and trained reviewers detect possible signs of serious safety risk, parents may be notified — but only with the information needed to support their teen’s safety. Parents can choose if they’d like to be contacted by email, SMS, push notification, or all three.
Parents can adjust certain features, set time limits, and add safeguards that work for their family including:
Set quiet hours, or specific times when ChatGPT can’t be used.Turn off voice mode, to remove the option to use voice mode in ChatGPT.Turn off memory, so ChatGPT won’t save and use memories when responding.Remove image generation, so ChatGPT won’t have the ability to create or edit images.Opt out of model training,so their teen’s conversations won’t be used to improve models powering ChatGPT and Sora.

The remaining Max users on our waitlist were granted access to Claude in Chrome, along with the following updates:
Powered by Sonnet 4.5:Claude in Chrome now defaults to Sonnet 4.5, our smartest model yet. Improved for browser tasks, you'll notice better reasoning, fewer errors, and more reliable task completion—especially for multi-step workflows.Work across multiple tabs:Claude can now juggle multiple browser tabs at once. Just drag tabs into Claude's tab group and it can see and work across all of them simultaneously—no more jumping back and forth to gather information before taking action.Smarter on the sites you use every day:Claude now understands how to navigate Slack, Google Calendar, Gmail, Google Docs, and GitHub without you having to direct every click. Ask Claude to "schedule a meeting" or "update the doc" and it knows what to do—no need to walk it through every step. We’re continuing to improve Claude’s understanding of other popular sites.Get notified when Claude needs you:Turn on notifications and Claude will ping you when it needs your permission or when a task is complete. Now you can switch to other work while Claude handles things in the background.Show Claude exactly what you mean:Upload images or take screenshots of specific screen areas to give Claude precise visual context. Much faster than describing what you're looking at.

- Released Gemini Robotics-ER 1.5 model in preview. See the
[Robotics overview](https://ai.google.dev/gemini-api/docs/robotics-overview)
to learn about how to use the model for your robotics application.

- Launched following preview models:

- `gemini-2.5-flash-preview-09-2025`
- `gemini-2.5-flash-lite-preview-09-2025`

See the [Models](https://ai.google.dev/gemini-api/docs/models) page for details.

We’re launching an early preview of ChatGPT pulse forPro users on iOS and Android. This is a new experience where ChatGPT can now proactively deliver personalized, focused updates for you once a day. These updates are based on asynchronous research ChatGPT can do on your behalf overnight, synthesizing information from your memory, chat history, and direct feedback to learn what’s most relevant to you.
Updates are delivered once a day as visual summaries you can scan at a glance or dive deeper into. Each topic opens into more detail where you can ask follow-up questions. You can shape what shows up in pulse with a quick thumbs up or down, or by using curate to tell ChatGPT what you’d like to see.
If you ask for something new in the evening, we aim to include it the next morning, though exact timing may vary. Pulse refreshes daily, and you can save any item into a chat to keep it.
As a preview, pulse has some limitations. It aims to show you what’s most relevant and useful but you may still see suggestions that miss the mark. For example, you may get tips for a project you already completed. You can guide what shows up by telling ChatGPT directly. It remembers your feedback for next time and improves as it learns from real use.
It requires saved memory and chat history to be turned on, and you can switch pulse off anytime in settings. SeeChatGPT Pulse.

Update to 2.5 Flash in GeminiWhat:An enhanced version of Gemini 2.5 Flash is now available to Gemini app users. Responses are now better organised with improved formatting. Gemini 2.5 Flash uses elements like headers, lists and tables to make even complex outputs clearer and easier to understand. We've also improved image understanding, so that Gemini can better interpret and respond to your uploads.To feel the difference, try working through a problem at school or work involving an image, like a diagram, and asking Gemini to walk you through it; or upload your class notes and ask Gemini to organise, summarise or create flashcards for you.Why:We're committed to continually improving performance and making Gemini a more insightful, collaborative partner. When it's better at understanding context and images, it's easier for you to create, interact and accomplish your goals.

- Released `gemini-2.5-flash-native-audio-preview-09-2025`, a new native audio model for the Live API with improved function calling and speech cut off handling. To learn more, see the [Live API guide](https://ai.google.dev/gemini-api/docs/live-guide) and [Gemini 2.5 Flash Native Audio](https://ai.google.dev/gemini-api/docs/models#gemini-2.5-flash-native-audio).

We're launching ChatGPT Go, our low cost subscription plan, in Indonesia.
For Rp 75.000/month, ChatGPT Go provides everything included in theFreeplan, and:
More messages, larger file uploads, and expanded image generationAccess to advanced data analysisLonger memory for more personalized responses
Available on web, mobile (iOS & Android), and desktop (macOS & Windows).
Learn more aboutChatGPT Go.

Users on Enterprise plans can leverage memory; Claude can now remember relevant context from your chats and generate a memory summary. Incognito chats give you a way to exclude conversations from Claude’s memory. Refer to the following articles for more information:
Using Claude’s chat search and memory to build on previous contextImporting and exporting your memory from ClaudeUsing incognito chats

Your custom Gems are now shareableWhat:Starting today, you can share the Gems that you create with friends, family and colleagues. Gems are a simple way to customise Gemini to achieve your specific goals. This means that you can tackle recurring tasks without retyping lengthy prompts and providing context each time.Share your weekly meal planner Gem to help a friend get inspiration, or send your holiday planner Gem to a family member before their big trip. To share, simply open your Gem manager on the web and click 'Share' next to any Gem that you've created. Sharing Gems works just like sharing files in Google Drive, giving you control over who can view or edit your Gem. Visit theHelp Centreto learn more about sharing Gems.Why:Turn your favourite Gems into a shared resource so that you can help others also save time on repetitive prompting and get started more easily. Rather than creating from scratch, you can use a fine-tuned Gem with effective instructions with friends or even an entire organisation.