Google Gemini
Google Gemini
Tools & services
Overview
Gemini from Google is a generative AI tool and conversational assistant capable of processing text, code, audio, images, and video for a wide range of tasks.
Harvard Google Gemini licenses include access to:
Basic tier (available to all Harvard Google Workspace users):
- Gemini Chat: Conversational AI for general questions and search, brainstorming, summarization and editing, translation, data analysis and general text / code generation.
- Source-grounded research (Gemini Notebook): Upload up to 300 sources (up to 500MB/500K words per source) with up to 500 daily chat prompts to generate summaries, audio/video overviews, quizzes, and flashcards grounded strictly in your source documents.
- Web-based deep research: Live web integration for synthesized web searches, reference gathering, and research summaries.
- Custom assistants (Gems) and Gemini Canvas: Create, use, and share customized prompt-driven assistants (Gems) and interactive workspaces (Gemini Canvas) for coding, docs, study guides, and presentations.
- Image and multimedia generation: Create text, images, short audio clips, and short video files.
- Google Workspace extensions: Direct integration across standard core Google tools, including Docs, Drive, Slides, Gmail, and Forms.
Premium tier (by request at additional cost):
- Gemini Advanced / Gemini for Google Workspace: Includes access to higher-capacity models like Gemini Pro, expanded context windows, custom Gems, and integration across Docs, Drive, Gmail, and Slides.
Access to connectors and plugins: Gemini can integrate with other tools and services via “connectors” to read data and perform tasks. All connectors must be reviewed and approved for security and privacy compliance before being enabled. View a full list of connectors that have been approved for use with Harvard Gemini.
Not included
- API keys: If you need an API key for programmatic use, request access to the Gemini API through Harvard's API Platform.
- Google AI Studio and third-party extensions requiring external data sharing are not included.
Limitations
- Subject to model rate limits and context length constraints per session.
Security
- Standard Products are approved for work using up to Level 3 confidential data.
- Agent is approved for tasks involving up to Level 3 confidential data providing you actively monitor autonomous sessions.
Privacy
- Data you enter will not be used to train Google models.
- Harvard cannot access data from user sessions.
Request a license
- All Harvard Google users have access to the basic tier of Gemini. Sign in with your HarvardKey at gemini.google.com.
- To request a premium license, use the form linked below.