AI models
Notes on specific AI models (chat assistants, language models, image models, and open-weight models) and when each is a good fit. This is where colleagues share which model they reached for and how it went. Ordered alphabetically by title.
OpenAI (ChatGPT)
GPT-models (Unknown version)
Modality: Text, Code · Purpose: Coding, Learning Aid, Productivity, Research, Writing
Several 2026 Upskilling Fellows reached for it as a general-purpose research, writing, and learning partner across very different fields. Avaliable via AIStudio.
Reviews:
- 07/2026 — Wangda Zuo — Recommend. Useful as a research and writing partner for reports, proposals, and keynote presentations, and for discussing and refining ideas.
- 07/2026 — Wenrui Hao — Recommend. Helpful for learning unfamiliar foundational concepts (used alongside the IBM Qiskit learning ecosystem) and for asking basic questions on a fast-changing, technically difficult topic.
- 07/2026 — Jia Li — Recommend. Valuable as an on-demand learning companion for navigating inconsistent or quickly outdated texts and connecting mathematical and intuitive explanations.
- 07/2026 — Justin Silverman — Recommend. Used to generate Markdown-formatted flashcards as part of a spaced-repetition study workflow; see the Anki flashcards entry in Tutorials and guides.
Source: 2026 Summer Faculty Upskilling Reflection Lunch meeting notes.
Anthropic/Claude
Multiple 2026 Upskilling Fellows reached for Claude as a general-purpose research, writing, and learning partner across very different fields. Haiku and Sonnet are avaliable via AIStudio. Fellows requested that Penn State make Opus and Fable avaliable under Penn State accounts, even if there are limits and/or costs.
Fable
Modality: Text, Code · Purpose: Coding, Learning Aid, Productivity, Research, Writing
Reviews:
- 07/2026 — Eric Ford — Recommend. Widely recognized as one of most powerful models at the time. Higher cost and usage limits mean that we should probably reserve it for particularly challenging tasks or tasks where we’re not satisfied with performance of other models.
Opus
Modality: Text, Code · Purpose: Coding, Learning Aid, Productivity, Research, Writing
Reviews:
- 07/2026 — Eric Ford — Recommend. Good for analyzing sizable codes, making plans to refactor, add features, etc.
Sonnet
Modality: Text, Code · Purpose: Coding, Critical Feedback, Learning Aid, Productivity, Research, Writing
Reviews:
- 07/2026 — Eric Ford — Highly Recommend. Effective for implementing code changes, especially if provided a plan (e.g., from Opus or Fable).
Haiku
Modality: Text· Purpose: Productivity, Writing
Multiple 2026 Upskilling Fellows reached for it as a general-purpose research, writing, and learning partner across very different fields. Haiku and Sonnet are avaliable via AIStudio.
Reviews:
- 07/2026 — Eric Ford — Recommend. Good for summarizing text, categorizing responses.
Penn State AI Services: model and tool overview
Level: Beginner · Modality: Text, Code, Images · Purpose: Research, Productivity
Penn State’s own overview of the AI services and models available to faculty and staff, including which are cleared for different kinds of university data. Start here before choosing a model for real work.
Reviews:
- 07/2026 — Penn State — Highly Recommend. Choosing a model requires attention to data policy as well as capability.
To add an entry, copy the block in templates/resource-entry-template.md. See CONTRIBUTING.md.
Before using any AI model for university work, check what is approved for your data. Penn State’s AI Comparison Table and AI Tools: Help Me Choose are the authoritative guides on approved tools and data-sensitivity levels.