GenAI Tool Features | AI Competency Centre
GenAI Tool Features
Features last updated 13 November 2025
The University of Oxford is supporting various generative AI tools for use by staff and students, including ChatGPT Edu, Google Gemini, and Microsoft Copilot. These tools have differing features accessible through University accounts. This page states which features you should expect to see when you access supported tools via your SSO-linked accounts.
ChatGPT Edu
When logging in with an SSO, all staff and students will join the University of Oxford ChatGPT Edu workspace. The features available in the University of Oxford workspace will be subject to internal governance based on business need and may not be the same as those in consumer versions. This guide outlines what is available.
ChatGPT’s advanced features (5.1-Thinking, Codex, Voice Mode, Video, Tasks and Image Generation) use significant shared resources. To enable inclusive access for all staff and students and manage access fairly across the whole University, usage limits for individual accounts will be introduced, with options to purchase additional access coming soon.
For guidance on getting started with ChatGPT Edu, see our ChatGPT Edu Onboarding Guide.
For video overviews of some of the features listed on this page, see the Canvas Module, which includes resources created by OpenAI.
For more details about your account and to migrate between accounts associated with different email addresses see this FAQ.
ChatGPT models:
These models are available to all staff and students.
- GPT-5.1
- GPT-5.1 Thinking (advanced feature)
- GPT-4o (legacy)
- GPT-5-mini
- GPT-5 Thinking mini (automatically triggered by GPT-5.1 for more difficult tasks but cannot be selected by the user)
Specific tools within ChatGPT Edu:
You can select these tools in the drop down menu or by typing / in the chat but you can also trigger them by asking in the prompt (for example, "generate an image of" or "search the web for..." or "use Canvas").
- Image generation (advanced feature)
- Web search
- Canvas (distinct from Oxford’s Canvas VLE)
The model powering ChatGPT Edu can also call tools that cannot be selected in the interface but can be triggered by asking in the prompt. They are:
- Memories - you can say “remember this about me” or “what do you remember about me?” or “forget this memory” - sometimes the model will choose to call the memories into its context when it think it may be relevant
- Code Interpreter - this is when the model writes and runs computer code to generate part of the answer (for instance, when you upload a spreadsheet and ask the model to analyze the data). Note: You can enable the Code Interpreter tool as a capability in custom GPTs.
Modes of interaction available to the user
- Dictate - speak about what you want included in the prompt (represented by the microphone symbol)
- Voice mode - have a live conversation (represented by the soundwaves symbol) | (advanced feature)
- Video (within voice mode on mobile app only) | (advanced feature)
- Study mode (this mode changes the behaviour of the model and is equivalent to changing the prompt)
Major ChatGPT Edu features
We are aware of some limited permissions issues with Microsoft connectors affecting a number of users. We are currently working with the Nexus365 team to restore full functionality.
Connectors let ChatGPT securely connect to third-party apps, such as Outlook and SharePoint so that you can search files, pull live data, and reference content directly in your ChatGPT chats.
In October 2025, this feature was expanded and rebranded as Company Knowledge, which provides users with the option for ChatGPT to automatically select the appropriate source across a range of previously individually selected sources (SharePoint, Outlook Email, Teams).
Despite its name, granting access to Connectors and Company Knowledge does not grant access to all company files. Individuals must still decide whether they wish to enable a connector, and once enabled, it can only access files they already have permission to view.
Available Connectors
Connectors available to staff at the University will appear in black text within Settings > Connectors. Connectors not yet available will appear greyed out.
To enable a connector, select it from the Connector menu and connect your personal Microsoft or GitHub account.
The current connectors available to staff are:
- Outlook calendar
- Outlook email
- SharePoint
- GitHub
How to Use Connectors
Once connected, your connectors will appear in the tool selection menu as ‘Company Knowledge’. You can ask ChatGPT to search files, fetch live data, or reference documents by asking it to use your Company Knowledge, or by manually selecting and enabling this tool in Settings.
Remember: Despite its title Company Knowledge, Connectors can only access files that you have permission to access through your connected accounts.
Projects are smart workspaces that keep everything related to a long‑running effort in one place. You can group together chats, upload reference files, and add custom instructions so ChatGPT remembers what matters and stays on‑topic. With memory, context, and flexible tools, they’re ideal for repeated and evolving work such as writing, research, planning, and more.
Shared projects are not currently enabled in the University's workspace.
View a short video overview of Projects created by OpenAI.
With a SSO-linked ChatGPT Edu account, you can:
- Create and manage custom GPTs
- Share custom GPTs to the University's workspace and externally (please refer to guidance on sharing custom GPTs)
- Access custom GPTs shared in the University's workspace (this does not apply to third-party GPTs due to data concerns)
We have created an Oxford GPT Library to showcase some of the custom GPTs created and shared within the University's workspace.
View a short video overview of Custom GPTs created by OpenAI.
Tasks allow you to schedule actions that ChatGPT can do on your behalf (for example, summarise the week’s top news from a page, create a daily French lesson).
To schedule a task, simply ask ChatGPT in a chat that you want to do this - describe the task and the frequency. (Please Note: You must be using the GPT-5 Thinking model to create a Task.) You can schedule one-time or recurring tasks. Each task runs automatically based on the schedule you define. You can view, edit, or cancel your active tasks from the Tasks page.
Tasks are currently supported on ChatGPT Web, iOS, Android, and macOS. They are not supported on the Windows Desktop app. The Schedules page, which is used to view and manage all tasks, is currently available only on ChatGPT Web.
Codex is a coding agent using GPT models that can be accessed via the ChatGPT Edu login but is not part of ChatGPT itself.
Codex covers several different capabilities and the ones currently enabled are as below:
- Codex Cloud / Web Codex Cloud is accessible at https://chatgpt.com/codex or by clicking the Codex button in the left sidebar when on https://chatgpt.com. Using Codex Cloud, users can connect to a GitHub repository, and ask in natural language to make changes. Codex Cloud will not make changes directly to the codebase but let the users create create pull requests with suggested changes in the associated GitHub repository.
- Codex CLI This is a coding agent that software engineers can access through the terminal using a command line interface. Users can download the tool from the link, and login using their ChatGPT Edu account. This still uses natural language interactions to allow users to make changes to code. Unlike the Codex Cloud, you do not have to explicitly connect to a GitHub repository. This tool has access to your local files, and can run commands directly on your computer.
- Codex IDE This is similar to Codex CLI except that this is an extension directly for an IDE program. Currently OpenAI offer Codex IDE extensions for three popular IDEs, Visual Studio Code, Cursor, and Windsurf.