LLM Changelog

We’ve made further improvements to search in ChatGPT, making results more accurate, reliable, and useful.
Factuality:Fewer hallucinations, improving answer quality.Shopping:Search is better at detecting shopping intent, showing products when you want them and keeping results focused when you don’t.Formatting: Answers are better formatted for quick understanding without losing detail or quality.

We’re adding GPT-5-codex, a GPT-5 variant optimized for agentic coding in Codex. It’s available everywhere you use Codex: default for cloud tasks and code review, and selectable for local workflows via the Codex CLI and IDE extension. Use GPT-5-codex for coding-focused work in Codex, or Codex-like environments; use GPT-5 for general, non-coding tasks. Please review theannouncementblog for more information.
Note: GPT-5-Codex is not currently supported in ChatGPT or the API.
To learn more about Codex, visit thedevelopers siteas well as our general help article:Using Codex with your ChatGPT plan.

Last month we announced that everyone now has access to Advanced Voice Mode, with usage expanded from minutes per day to hours for Free users, and near-unlimited use for Plus.
We also shared that Standard Voice Mode would be retired following a 30-day sunset period. We’ve heard feedback that Standard Voice is special to some folks, and we want to make sure the transition feels right. To that end, we’re keeping Standard Voice available while we address some feedback from its users about Advanced Voice.
Stay tuned—more improvements to Voice are coming soon.

You can now branch conversations in ChatGPT, making it easier to explore different directions without losing your original thread. Hover over a message, clickMore actions(⋯), and selectBranch in new chatto start a separate conversation from that point.
This feature is available today for all logged-in users on web.

Projects bring your chats and files into one place, helping you stay focused and get things done. They’re now available on our Free tier, with up to 5 file uploads per project.
We’ve also updated file uploads on paid tiers: Plus, Go, and Edu support up to 25 files per project. Pro, Business, and Enterprise support up to 40 files per project.
New customization options are available for all plans, including colors and icons, to help keep your projects organized.
Learn more aboutprojects in ChatGPT.

Students get one year of Google AI Pro at no cost in the UK, Germany, Egypt, Saudi Arabia and Mexico.What:Google is offering students aged 18 and older in the UK, Germany, Egypt, Saudi Arabia, and Mexico a one-year upgrade to Google AI Pro at no cost. Students must sign up for this promotion by 3 November 2025. This plan includes unlimited chat, image uploads and quiz generations for homework help and exam prep, plus 2 TB of storage. Plan automatically renews at £18.99 / month after trial. Cancel anytime.Terms apply.Students will have access to our Gemini 2.5 Pro model, Veo 3 Fast, Deep Research and an enhanced NotebookLM with 5x more notebooks and audio overviews.Why:Our mission is to empower every student with the AI tools and skills that they need to advance their learning, unlock their creativity and build the future of work.

Starting today, Codex works with you everywhere you build—in your terminal or IDE, on the web, in GitHub, and even from the ChatGPT iOS app. Your ChatGPT account connects it all, so you can work seamlessly between your local environment and Codex’s cloud without losing state.
We’re excited to introduce the latest Codex updates:
IDE Extension:The new extension brings codex into VS Code, Cursor, and other VS Code forks, so that you can seamlessly preview local changes and edit codeSign in with ChatGPT:Available in both the IDE and CLI, eliminating API key setup and providing access directly through your existing ChatGPT planSeamless Local ↔ Cloud Handoff:Developers can pair with Codex locally and then delegate tasks to the cloud to execute asynchronously without losing stateUpgraded Codex CLI:Refreshed UI, new commands, and bug fixesCode reviews in GitHub:Set up Codex to automatically review new PRs in a repo, or mention @codex in PRs to get reviews and suggested fixes
Additionally, all product information and updates for Codex moving forward will be announced on our new site:developers.openai.com/codex.
We invite you to explore the site for more details on these new features, as well as guides on how to get started.
To learn more about Codex, visit the newdevelopers siteas well as our general help article:Using Codex with your ChatGPT plan.

Reimagine your photos with a promptWhat:Experience a new level of creativity with our latest image generation and editing model. We've improved how Gemini follows your instructions, making it easier to get the results that you want. Upload multiple images to combine concepts, borrow creative elements or blend scenes to create something unique. Plus, when you use your own selfie, Gemini does a better job of keeping your look consistent, so that you can star in any adventure you can imagine.Why:We believe that the best creative tools are the ones that get out of your way and let your imagination run wild. This update is a step forward in giving you more precise control and powerful new ways to express your ideas. We can't wait to see what you create!

An improvement toprojectsis now available. When creating a project, users have the option to enableproject-only memory.
Withproject-only memoryenabled, ChatGPT can use other conversations in that project for additional context, and won’t use yoursaved memoriesfrom outside the project to shape responses. Additionally, it won’t carry anything from the project into future chats outside of the project.
This creates a focused, self-contained space, which is useful for long-running or sensitive work where you want ChatGPT to stay anchored to that project’s tone, context, and history.
Note:
Personal Memorymust be enabled to utilize this feature.Settings->Personalization->MemoryThis feature will initially only be available on the ChatGPT website and Windows app. Support for mobile (iOS and Android) and macOS app will follow in the coming weeks.
Learn more aboutmemory in projects.

Revisit your conversations: You can now search your Gemini chat history.What:Find and revisit past conversations in Gemini. To get started, simply expand your navigation menu and tap the magnifying glass or the “Search for chats” bar. Enter a keyword or phrase, and Gemini will display all relevant conversations. This feature is live globally on web and mobile.Why:This is one of our most highly requested features, and lets you collaborate more seamlessly with Gemini. No matter if you're working on a project from last week or searching for a prompt you crafted a month ago, the search bar helps you find what you need in seconds to keep your momentum going.

- Released [URL context tool](https://ai.google.dev/gemini-api/docs/url-context) to general availability (GA), a tool for providing URLs as additional context to prompts. Support for using URL context with the `gemini-2.0-flash` model (available during experimental release) will be discontinued in one week.

We’re launching ChatGPT Go, a new low-cost subscription plan. This will initially be available for users in India only.
For ₹399/month (GST included), ChatGPT Go provides everything included in theFree plan, 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).
ChatGPT Go is geo-restricted to India at launch, and is able to be subscribed to by credit card or UPI.
Learn more aboutChatGPT Go.

We’re making GPT-5’s default personality warmer and more familiar. This is in response to user feedback that the initial version of GPT-5 came across as too reserved and professional. The differences in personality should feel subtle but create a noticeably more approachable ChatGPT experience.
Warmth here means small acknowledgements that make interactions feel more personable — for example, “Good question,” “Great start,” or briefly recognizing the user’s circumstances when relevant. This is different from sycophancy, which we define as excessive flattery that can feel insincere. You can find more about our learnings on reducing sycophancyhere.
This update does not show an increase in sycophancy compared to the previous personality of GPT-5, measured by the same internal evals we ran and detailed in oursystem card. Teaching models to be warm without being sycophantic is an ongoing research challenge, and we will continue iterating to get this right. More updates are coming soon.

Introducing Temporary ChatWhat:Now you can have a quick, one-off conversation with the Gemini app that won’t be saved to your chat history. Our new Temporary Chat feature is great for when you're exploring private questions or brainstorming an idea that's outside your usual style. To start, select the “Temporary Chat” icon next to “New Chat”. This feature will be rolling out over the coming weeks, starting today.Why:With Temporary Chat, you can have conversations that won’t appear in your recent chats or Gemini Apps Activity. These chats won’t be used to train Google’s AI models or personalize your Gemini experience, giving you a clean slate whenever you need one.

You can now choose between “Auto”, “Fast”, and “Thinking” for GPT-5. Most users will want Auto, but the additional control will be useful for some people.
ChatGPT Plus users now have 3,000 messages/week with GPT-5 Thinking, and then extra capacity on GPT-5 Thinking mini after that limit. The context limit for GPT-5 Thinking is 196k tokens. We may have to update rate limits over time depending on usage.
4o is back in the model picker for all paid users by default. Paid users also now have a “Show additional models” toggle in ChatGPT web settings which will add models like o3, o4-mini, 4.1, and GPT-5 Thinking mini. 4.5 is only available to Pro users due to GPUs.

Gmail, Google Calendar, and Google Contacts are now available to connect and use in chat. Once you enable them, ChatGPT will automatically reference them when relevant, making it faster and easier to bring information from these tools into your conversations without having to manually select them each time.
This capability is part of GPT-5 and will begin rolling out to Pro users globally this week, followed by Plus, Team, Enterprise, and Edu plans in the coming weeks. To enable, visitSettings→Connectors→Connecton the application.
If you already have Gmail or Google Calendar enabled for deep research, you can now also use them in chat. To use them in deep research, you will still need to enable each connector separately and select it every time you start a new deep research request.
Learn more aboutconnectors.

Now connect even more tools in ChatGPT for useful, relevant responses in chat.
The following connectors are now available to search by chat in addition to deep research.
Plus:Box, Canva, Dropbox, HubSpot, Notion, Microsoft SharePoint, and Microsoft TeamsPro: Microsoft Teams and Github
Rollout begins today, and all eligible accounts will have access in the coming days.
To enable, visitSettings→Connectors→Connecton the application. Note that connectors are still in beta and defaulted off for Enterprise & Edu plans. Admins can enable them for their workspace inSettings.
Note:Connectors for Plus/Pro plans are not available in EEA, Switzerland, and the UK.

GPT-5 is slowly rolling out to all users on ChatGPT Plus, Pro, Team, and Free plans worldwide across web, mobile, and desktop. GPT-5 will be available to ChatGPT Enterprise and Edu plans soon.
GPT-5 in ChatGPT is our next flagship model and the new default for all logged-in users. It simplifies ChatGPT to a single auto-switching system that brings together the best of our previous models into asmart, fast model.
GPT-5 is available to all ChatGPT Tiers. Users on Paid tiers - Plus, Pro, and Team - have access to the model picker, which enables you to manually select GPT-5 or GPT-5 Thinking. Pro and Team tier users have access to GPT-5 Thinking Pro, which takes a bit longer to think but delivers the accuracy you need for complex tasks.
Learn more about GPT-5 in ChatGPT.

You can now choose from four distinct personalities or use the Default personality in yourCustomize ChatGPT settings. Default is the standard ChatGPT style: clear, neutral, and adaptable. The other personalities each have their own style and tone, described below.
Cynic– Sarcastic and dry, delivers blunt help with wit. Often teases, but provides direct, practical answers when it matters.
Robot– Precise, efficient, and emotionless, delivering direct answers without extra words.
Listener– Warm and laid-back, reflecting your thoughts back with calm clarity and light wit.
Nerd– Playful and curious, explaining concepts clearly while celebrating knowledge and discovery.
Please note that these personalities will not apply to Voice mode.

You can now set an accent color that applies to elements in ChatGPT, including your conversation bubbles, the Voice button, and highlighted text.
On web:Click your profile icon at the bottom left, selectSettings, go to theGeneraltab, and choose an option from theAccent colordrop-down.
On mobile (iOS and Android):Tap your profile icon at the bottom, go toPersonalization, then selectColor Schemeto pick your accent color.

Today we’re rolling out improvements to Voice Mode to make it more accessible and useful for everyone. It also now works with custom GPTs. We’re expanding access with near-unlimited use for Plus users and hours each day for Free users. To simplify the experience, Standard Voice Mode will be retired in 30 days, unifying all users onto our latest voice experience.
For paid users, Voice now adapts to your instructions, adjusting its speaking style (length, speed, tone, and more) to fit the moment.

Study effectively with helpful exam prep toolsWhat:Dive deeper into your studies by generating unlimited custom quizzes. You can also create flashcards and study guides based on your quiz, or other class materials, for an easy way to review and reinforce your learning. This experience is open to all users over the age of 18.Why:Effective studying is not one-size-fits-all. We want to provide a variety of ways for learners to master the topics they are studying, whether that is sitting down for a long, challenging quiz, quickly reviewing concepts with flashcards, or taking a comprehensive study guide with them on the go.Understand Complex Topics Faster with Integrated Visuals and YouTube videosWhat:Gemini now provides a richer learning experience by automatically integrating high-quality images, diagrams, and YouTube videos directly into its responses. When you ask about complex topics like photosynthesis or the parts of a cell, Gemini will seamlessly weave in visuals and YouTube videos alongside the text to help you understand the information more easily. Integration of high quality images, diagrams and YouTube videos is available to all users.Why:We believe seeing a concept is key to truly understanding it. Gemini now uses visuals to make complex information more digestible and memorable. This helps turn abstract ideas into concrete knowledge, making learning faster and more effective for everyone.Go Beyond the Answer with Step-by-Step Guided LearningWhat:Gemini now offers Guided Learning, a new mode designed to help you build a deeper understanding of whatever you’re trying to learn or do. Instead of providing a single answer, this feature guides you through subjects step-by-step, breaking down concepts and providing interactive help along the way. It's ideal for understanding complex information or developing a new skill. You can activate this tool using the ‘Guided Learning’ chip on desktop or the ‘Learn’ chip on mobile. This experience is open to all users.Why:

Solve your most complex problems with Gemini 2.5 Deep ThinkWhat:Starting today, Google AI Ultra subscribers get early, limited access to Gemini 2.5 Deep Think, Gemini’s most advanced reasoning mode. Deep Think is capable of thinking for longer and generating multiple parallel streams of thought simultaneously—much like how humans brainstorm to tackle complex problems – making it excel at iterative development and design, scientific and mathematical research, and coding.Try it next time you need more brain power for your most complex tasks, especially those requiring creativity, strategic planning and step-by-step improvements. Select 2.5 Pro from the model drop down, and "Deep Think" will appear in the prompt bar. Submit your task with Deep Think enabled and Gemini will let you know when your response is ready – generally in a few minutes.Why:We're committed to bringing Google's latest AI innovations faster to Google AI Ultra users, including our most powerful reasoning capabilities to help you tackle even the most complex problems.

Study mode is currently available to users in Free, Plus, Pro, and Teams plans globally and will expand to Edu plans in the coming weeks. Study mode works with any model available in ChatGPT on iOS, Android, web, and desktop.
Study mode is a new learning experience in ChatGPT designed to help you build a deeper understanding of any topic. When you turn it on, ChatGPT will ask interactive questions to understand your goals and skill level, then work with you to reach the answer together.
With Study mode enabled, ChatGPT can:
Guide understanding with Socratic-style questions.Break concepts into easy to follow sections starting simple and adding complexity as you progress.Personalize responses based on your past chats if memory is on, using examples and tips tailored to you.Check your understanding with open-ended prompts and feedback.Work with your materials by referencing images or PDFs you upload.
You can enable study mode at any time by selectingToolsin the prompt window and choosingStudy and learnfrom the drop‑down menu or go tochatgpt.com/studymode.
Study mode is powered by custom system instructions and can have some inconsistent behavior and mistakes across conversations. We plan on training this behavior directly into our main models once we’ve learned what works best through iteration and user feedback.
Learn more about study mode.

- Launched `veo-3.0-generate-preview`, the latest update to Veo introducing
video with audio generation. To learn more about Veo 3, visit the [Veo](https://ai.google.dev/gemini-api/docs/video) page.

- Increased rate limits for Imagen 4 Standard and Ultra. Visit the
[Rate limits](https://ai.google.dev/gemini-api/docs/rate-limits) page for more details.

Record mode is now available to ChatGPT Plus users globally in the macOS desktop app.
Record live conversations like team meetings or voice notes, and turn those into an editable summary in canvas.Available on the macOS desktop app only.
Learn more aboutrecord mode.
Note: Record mode originally rolled out to Team users on June 4, and Enterprise & Edu plans on June 18.

Stay on top of your work with the Productivity Planner GemWhat:The new Productivity Planner Gem seamlessly brings together information from your favourite productivity apps like Gmail, Calendar and Drive to help you stay organised. Whether it's getting an overview of your urgent emails or your calendar events for the week, this pre-made Gem gives you useful context to start your day and prioritise your most important tasks. Get started by asking it to summarise your key projects for the week, create a daily work brief or recommend action items to focus on.Availability: The Gem is available to Google AI Pro and Ultra subscribers, as well as qualifying Google Workspace Business and Education plans. Google Workspace apps must be enabled for all users to access the Gem.Why:Productivity Planner Gem helps you save time by connecting to your Workspace apps like Gmail, Calendar and Drive to deliver insights right when you need them. Schedule updates so that you can prompt less while staying informed on the things that matter most to you.

Pro users are now able to use chat search connectors (in addition to deep research connectors) for the following integrations: Dropbox, Box, Google Drive (synced and non-synced), Microsoft OneDrive (Business), Microsoft SharePoint.
Note this feature is currently limited to users located outside of the EEA, Switzerland, and the UK.
Learn more aboutConnectors.

Plan ahead with scheduled actionsWhat:You can now schedule actions in the Gemini app so you can plan ahead for future and recurring tasks.Ask Gemini to help you get organized with a daily summary of your calendar, to-do's and important unread emails. Or get weekly creative ideas for a project and updates on topics you’re passionate about.You can schedule recurring actions, like finding local weekend music events every Friday, or one-time tasks, like asking for a recap of a keynote the next morning. Just tell Gemini what you need and when—it'll handle it for you.Scheduled actions are available to Google AI Pro and Ultra subscribers and qualifying Google Workspace business and education plans.Why:We want Gemini to be your personal, proactive and powerful AI assistant, and scheduled actions can help boost your productivity in new and exciting ways with proactive help that free up your time.

Capture meetings, brainstorms, or voice notes. Launching today for Pro, Enterprise, and Edu users. Previously launched for Team users on June 4, 2025.
ChatGPT will transcribe, summarize, and turn them into helpful outputs like follow-ups, plans, or even code.Available on the macOS desktop app only.
Learn more aboutChatGPT Record.

- Released `gemini-2.5-pro`, the stable version of our most powerful model, now with adaptive thinking. To learn more, see [Gemini 2.5 Pro](https://ai.google.dev/gemini-api/docs/models#gemini-2.5-pro) and [Thinking](https://ai.google.dev/gemini-api/docs/thinking). `gemini-2.5-pro-preview-05-06` will be redirected to `gemini-2.5-pro` on June 26, 2025.
- Released `gemini-2.5-flash`, our first stable 2.5 Flash model. To learn more, see [Gemini 2.5 Flash](https://ai.google.dev/gemini-api/docs/models#gemini-2.5-flash). `gemini-2.5-flash-preview-04-17` will be deprecated on July 15, 2025.
- Released `gemini-2.5-flash-lite-preview-06-17`, a low-cost, high-performance Gemini 2.5 model. To learn more, see [Gemini 2.5 Flash-Lite
Preview](https://ai.google.dev/gemini-api/docs/models#gemini-2.5-flash-lite).

Enterprise and Edu users can also now choose from the full set of ChatGPT models (GPT-4o, o3, o4-mini and more) when building Custom GPTs. This was launched to Plus, Pro and Team users earlier this month.
Key details:
GPTswithout Custom Actionscan use the model picker to select from all models available to the user.GPTswith Custom Actionscurrently support GPT-4o and 4.1Available onwebfor users on all paid plans (Plus, Pro Team, Enterprise, Edu).