As you might know, the world of large language models (LLMs) and GenAI is moving quickly—and the Stanford AI Playground (aiplayground.stanford.edu) is keeping pace.
The AI Playground is a Stanford-managed platform that provides access to various LLMs, giving our community a more secure way to learn and experiment with popular AI models and tools.
Since its launch in fall 2024, the AI Playground has added:
- 6,900+ users (across Stanford faculty, staff, students)
- 1.4+ million messages
- 35+ billion tokens
The AI Playground team continues to add more models and functionality almost monthly. Many of these advances are in response to requests from our community.
"The true power of AI comes from ensuring equitable access to AI technologies, so that everyone can benefit. The AI Playground provides a safe place for all Stanford faculty, staff, and students to explore, create, and solve problems with cutting edge AI tools."
– Joshua Barnett
Director, Enterprise Infrastructure Architecture, Stanford University IT
Then, now, and next
The AI Playground initially launched with many of the same standard features available in commercial GenAI services, such as an easy-to-use chat interface, chat history, and the ability to switch among LLMs. We started with early models of OpenAI ChatGPT, Google Gemini, and Anthropic Claude.
Now, we have more LLMs, including Deepseek and Meta’s Llama, along with newer versions of the early models. Plus, we have new plugins like DALL-E for image generation, Wolfram for computations and mathematics, and Google web search and Web Scraper to help enhance results.
Additionally, LibreChat—the open-source platform utilized for the AI Playground—has added several neat features, such as side-by-side comparisons for the same prompt and the ability to execute generated code directly in the browser. You can explore the current models and tools listed in the AI Playground Quick Start, or dive in and discover them firsthand in the AI Playground.
Soon, the AI Playground will have a plugin for exploring the Stanford Admin Guide. Plus, be on the lookout for a new API gateway service that will enable developers to build tools and applications powered by the AI Playground.
Why join now?
With the Stanford-hosted AI Playground, you can compare different commercially-available LLMs without having to sign up for each one separately.* You can also move information between LLMs from different vendors and test the results. And importantly, our Stanford-managed platform offers more privacy and security than the platforms as available on their own commercially.
Many Stanford faculty, staff, and students are using the AI Playground to increase comfort and develop new skills, allowing for new insights and solutions.
*The AI Playground doesn’t offer every feature available directly from a vendor, and you may consider paying or subscribing to an AI tool in certain cases (e.g., if you want AI integrated into another tool, such as Microsoft 365 Copilot).
How to play
The AI Playground team plans to continue improving the Playground with new models and features in the future. As new features are released, we encourage everyone: keep experimenting! You can find new ways to improve your results by selecting different models, adjusting the configuration options for those models, or simply rewording your prompts. For help with understanding prompts, try UIT’s GenAI Prompt Guide and GenAI Use Cases for Experimenting.
Please remember: The AI Playground is not approved for high-risk data. Be sure to adhere to Stanford’s information security and privacy policies when using the AI Playground.
Help and feedback
- To ask general questions and brainstorm with peers, join the Slack channel #ai-playground.
- For technical questions or support, send a Help request.
- Share your feedback with the project team by using the feedback form.
