What are Claude models?
The Harvard Chan AI Glossary
In our classes, we often pause to explain the ideas behind AI and the language used to describe it. These posts grew out of those conversations. They’re for anyone who would like to understand AI better, with no technical background assumed.
Claude is Anthropic’s family of AI models. A model is the trained component that interprets supplied information and produces responses. Claude model names identify members of that family, while the Claude application and Claude Code name products through which models can be used. Those are related but separate layers.
Names such as Opus, Sonnet, Haiku and Fable distinguish lines within the family. A version number identifies a particular release. The names help identify what is being selected; they do not establish an enduring order of superiority. Capabilities and availability should be checked for the particular version.
Models can differ in the kinds of information they accept, the complexity of work they handle, response speed and cost. Multimodal means able to work with more than one kind of information, such as text and images. A model’s support for a type of input does not establish how accurately it will interpret every example of it.
Consider asking for an explanation of a chart. The model needs access to the chart and instructions about the question. If it is asked to retrieve a spreadsheet or change a file, the application also needs connected tools—software functions that perform those operations. Choosing a Claude model does not automatically grant those connections or permissions.
Developers may access models through an API, an agreed interface for software to exchange requests and responses. A person using the Claude application encounters a more complete product with conversation, files and other available features. Results can differ because the surrounding information and tools differ, even when the underlying model is related.
The practical distinction is between a model’s capability and a product’s configured behavior. An assistant can sound knowledgeable without having retrieved a current source. A coding product can propose a change without proving that the changed program works. The model name helps identify one component of the arrangement; the task, supplied information and checks determine what the result establishes.
Source: Anthropic — Models Overview.
If you would like to find out more about AI and improve your leadership skills, consider the AI in Health Care Certificate of Specialization from the Advanced Learning Academy at Harvard T.H. Chan School of Public Health. You can take an individual course or work toward the full certificate.
AI is a fast-moving field. This explanation reflects our class discussions and related material at the time of writing. Terminology, capabilities and tools may change.
For suggested additions or corrections, please email me at tpanch@hsph.harvard.edu.
Dr. Trishan Panch
Program Director, AI in Health Care
Harvard T.H. Chan School of Public Health — Advanced Learning Academy