StanfordUniversity IT Responsible AI for everyone
AI CatalystBeta

An earlier week at Stanford

Week of October 5

to (earlier week)

Latest week

Release notes 4

What University IT changed in Stanford's AI services this week.

  1. Claude Sonnet 5.5 in the Stanford AI Playground

    Anthropic's Claude Sonnet 5.5 is now available in the Stanford AI Playground. It is another option for everyday tasks, coding and creating documents.

  2. Claude Sonnet 5.5 and GPT-Realtime-2.1 on the AI API Gateway (opens in a new tab)

    Anthropic's Claude Sonnet 5.5 and OpenAI's GPT-Realtime-2.1 and GPT-Realtime-2.1 Mini are now available through the AI API Gateway. The two OpenAI models are for applications built around realtime voice interactions. To add these or other new models to an existing key, use the Change or Cancel AI API Gateway Key Service form.

  3. GPT-6.1 Sol enabled in ChatGPT Edu

    GPT-6.1 Sol is now enabled in Stanford's ChatGPT Edu. The new model is another option for coding, reasoning and tasks that take several steps.

  4. ChatGPT Space now available in ChatGPT Edu (opens in a new tab)

    Space brings shared pages, files and project context together, so you can work with colleagues and ask ChatGPT or Codex to help develop and update your work. Collaborative slides and spreadsheets are coming later.

In the news 5

What the vendors and the wider AI world announced.

  1. Introducing Claude Sonnet 5.5 (opens in a new tab)

    Anthropic's faster, lower-cost Sonnet, strongest at everyday tasks, bug fixes and polished documents, now in the AI Playground and on the AI API Gateway.

    Anthropic

  2. Introducing GPT-6.1 Sol (opens in a new tab)

    OpenAI's upgrade to GPT-6 Sol for coding, computer use and multistep work; OpenAI says it shows in ChatGPT Work and Codex, not yet in regular chat.

    OpenAI

  3. OpenAI DevDay 2026 recap (opens in a new tab)

    More than 20 announcements across ChatGPT, Codex and the models, including GPT-6.1 Sol and Space, both now enabled in Stanford's ChatGPT Edu.

    OpenAI

  4. Higher Ed's AI Integration Overhyped, New Data Shows (opens in a new tab)

    Canvas data from 19.5 million users shows licensed AI tools rarely make it into courses, and only 11 percent of instructors have had thorough AI training.

    insidehighered.com

  5. Harvest Good Cyber Habits This October for Cybersecurity Awareness Month (opens in a new tab)

    October's security themes include AI-written phishing and faked voices or video: confirm unusual requests another way and keep high-risk data in approved tools.

    Stanford

Archive

Every published week