Body
Introduction
Claude for Education, part of the Generative AI @ Pitt offerings, gives you access to more than one version of Claude. Each version is called a "model," and each one is built for a different kind of work. Some are fast and light; others take more time to think through difficult problems.
Your model choice matters for two reasons. A model suited to your task gives you a better answer, and each model uses up your usage limit at a different rate. Choosing well means you can keep working longer before you reach a limit.
If you have not set up your account yet, start with Getting Started with Anthropic Claude for Education.
Information
Important: Anthropic releases new Claude models and retires older ones frequently, and a newly announced model may not be available at Pitt right away. The models listed below can change without notice. Pitt Digital will work to keep this article as up to date as possible. The model picker in Claude is always the accurate source for what is available to your account.
Available Claude Models
You can switch between these models at any time using the model picker near the message box.
| Model |
Best for |
Effect on your usage limit |
| Claude Haiku 4.5 |
Quick questions, short summaries, and pulling a specific detail out of a document |
Lightest |
| Claude Sonnet 5 |
Everyday work such as writing, editing, coding, research, and multi-step tasks. Start here if you are not sure which model to pick. |
Moderate |
| Claude Opus 5.5 |
Complex tasks: deep research, involved reasoning, and long or dense documents that need sustained analysis |
Heavy |
Two other things you may notice in the model picker:
- Selecting More models shows older versions, currently Opus 5, Opus 4.8, Opus 4.7, Opus 4.6, Opus 3, and Sonnet 4.6. These remain available for anyone who needs a specific earlier version, but the current models in the table above are generally the better choice.
- Claude's Fable models are part of a premium tier that is not included in the University's Claude for Education license, so they do not appear in your model picker.
Match the Model to the Task
These examples are a starting point rather than a rule. If a lighter model gives you everything you needed, stay with it.
For Students
| What you are doing |
Try this model |
| Condensing assigned reading or lecture notes into a study sheet |
Haiku 4.5 |
| Checking a citation format or looking up a term from class |
Haiku 4.5 |
| Getting feedback on an essay draft or a presentation outline |
Sonnet 5 |
| Being quizzed on a unit before an exam |
Sonnet 5 |
| Synthesizing several research papers for a literature review |
Opus 5.5 |
For Faculty
| What you are doing |
Try this model |
| Pulling key dates out of the academic calendar or a policy page |
Haiku 4.5 |
| Drafting a syllabus section, assignment prompt, or rubric |
Sonnet 5 |
| Turning lecture notes into slides or discussion questions |
Sonnet 5 |
| Finding themes in open-ended course evaluation comments |
Sonnet 5 |
| Reviewing a methodology section or drafting a grant narrative |
Opus 5.5 |
For Staff
| What you are doing |
Try this model |
| Summarizing a long email thread or a set of meeting notes |
Haiku 4.5 |
| Reformatting a list or pulling specific fields out of a document |
Haiku 4.5 |
| Writing an announcement, procedure, or training material |
Sonnet 5 |
| Building a spreadsheet formula or tidying up a data export |
Sonnet 5 |
| Comparing several vendor proposals against a list of requirements |
Opus 5.5 |
Understanding the Effort Setting
Along with choosing a model, you can choose how hard Claude works on your request. Select Effort in the model picker to see the levels. Higher effort means a more thorough response, but Claude takes longer to answer and uses your limit faster. Claude marks the default level in the menu, and not every model offers this control.
| Effort level |
What it does to the result |
Effect on your usage limit |
| Low |
Claude answers with very little deliberation. Fine for lookups, formatting, and short summaries, but it may miss nuance in a harder question. |
Lightest |
| Medium |
Some deliberation before answering. Suits straightforward requests that still need a bit of care. |
Light |
| High |
Thorough reasoning that fits the large majority of work. Stay here unless you have a reason to move. |
Moderate |
| Extra |
Deeper reasoning for genuinely difficult problems. Answers take noticeably longer to arrive. |
Heavy |
| Max |
The most thorough reasoning available, for the hardest problems you have. |
Heaviest. Claude notes in the menu that this level uses 1.5 times or more of your usage. |
Effort and model choice multiply each other. The heaviest model at the highest effort will consume your limit far faster than the same question asked of a lighter model at the default, so it is worth reserving that combination for the work that truly needs it.
Get More Out of Your Usage Limit
Claude measures your activity in units called tokens, and your account has a usage limit that refreshes periodically. Claude warns you as you approach it. These habits help that limit stretch further:
- Start with the lightest model that can do the job. If the answer covers everything you needed, you are done. If it misses something important, ask again with the next model up.
- Lower the Effort setting for routine work and raise it only when a harder problem calls for it.
- Reserve the heavier models for work that earns it. Using Opus for a task Haiku could handle spends more of your limit without improving the result.
- Start a new chat when you change topics. Claude re-reads the whole conversation each time you send a message, so a long chat about many subjects uses more of your limit than several focused chats.
Clear, specific requests also cut down on repeat attempts. See the Tips for Using Claude section of Getting Started with Anthropic Claude for Education.
Related Information