New AI API offerings available through Harvard’s API Portal

In addition to the current suite of AI tools that HUIT offers, new API-based solutions are now available to help developers and researchers integrate AI into their work more efficiently. APIs are a scalable and practical option for embedding AI into systems, scripts, or applications. 

These expanded AI API offerings streamline access and eliminate the need to set up individual accounts with providers and handle payments and reimbursements. AI APIs can be accessed securely through the API Portal (HarvardKey sign-in required) in the following ways, depending on your School/Unit affiliation: 

• Pay-as-you-go: Pay for your usage as you go; requires a HUIT Billing ID.  
  • AWS Bedrock API for interactions with the Anthropic, Meta, and Mistral models.
  • Google Gemini API for interactions with Flash 2.5 Pro, Flash 2.0 lite, embedding-005, and all other Gemini models.
  • OpenAI Direct API for interactions with all of the OpenAI models.
  • OpenAI via Azure for interactions with the OpenAI models hosted in the HUIT Azure account.  
• Credit-based access: Work within a predetermined set of OpenAI credits; limited availability.
  • OpenAI “Community Developer" API: $10/month in OpenAI API credits for students, faculty, staff, and researchers in FAS and Harvard College.
  • OpenAI Credit Redemption APIs for approved administrative, research, and teaching & learning use cases in FAS, SEAS, and Central Administration. 

If you are considering using Generative AI tools for Harvard work, you must follow the University’s initial guidelines for use. For additional guidance on these and other AI offerings, request a consultation with HUIT.