Skip to main content

Tools

CAIL Sandbox interface showing CUNY AI Lab chat platform
Open-weight models — Publicly available weights for transparency and auditability
Open-source platform — Built on Open WebUI, a community-driven chat interface for running AI models
Zero-retention — Prompts and outputs are never stored or used for vendor training
Citable configs — Shareable, reproducible configurations for scholarly use
CO₂ estimates — Visible energy and carbon cost per interaction
Accessibility — Text-to-speech, speech-to-text, and more planned

CAIL Sandbox

The CAIL Sandbox provides faculty, staff, and students with access to an AI chat interface through which they can access a curated set of open-weight models. CUNY faculty, students, and researchers can use the CAIL Sandbox to access the most robust open-weight AI models while ensuring that their chats and data remain secure. Our zero-retention providers (AWS Bedrock and OpenRouter) do not store any data from prompts or responses, and users can configure the models to their needs while also gaining insight into the environmental impacts of their tool use.

Access to the CAIL Sandbox is limited during the Spring 2026 semester. To learn more, contact us.

For a full list of available models with capabilities, licensing, and recommended use cases, see our Model Registry.

Launch CAIL Sandbox View Model Registry

Classroom Chatbots

Custom assistants built inside the CAIL Sandbox

AmigAI

A Spanish-language chatbot that helps CUNY students practice conversational Spanish through culturally responsive pedagogy.

Course-Specific Assistants

Built for individual courses using your syllabus, readings, and materials. Support students through complex texts with guiding questions.

Examples: Writing Process Assistant, Paratext Analysis Tool, Active Reading Companion

Project Templates

Reusable configurations designed for common classroom workflows that instructors can adapt across sections or semesters.