AI Playground Release Notes

December 2025

Dec. 17, 2025

Enabled the OpenAI GPT 5.2 model for all users.

Dec. 16, 2025

AI API Gateway - Upgraded the version of LiteLLM supporting virtual API access to the models within the AI Playground.

Dec. 12, 2025

Enabled the FinGate Financial Info Navigator (FIN) agent for all users.

Dec. 04, 2025

Changed configuration of agents menu to resolve an unintended visibility issue in production.

November 2025

Nov. 28, 2025

Deployed major new UI enhancements for a streamlined user experience.
Deployed several new features including:
- Smart user memories for conversational context
- New web search, files search, and improved rendering options in the prompt bar
- User feedback system to help further improve AI responses
- Restored the conversational search functions
- Enhanced OCR support using Mistral’s OCR API to assist in converting files to text automatically
- Additional LaTeX typesetting improvements for mathematical and technical conversations.
- More Performance & Reliability Improvements
Removed ScholarAI agent for all users. ScholarAI was purchased by a third party and was no longer able to support the security and privacy requirements outlined in the university agreement.

October 2025

Oct. 18, 2025

Enabled UIT Service Desk Search agent for users in the UIT Service Desk.

Oct. 04, 2025

Enabled the Claude 4.5 Sonnet model for all users.
AI API Gateway - Upgraded the version of LiteLLM supporting virtual API access to the models within the AI Playground.

September 2025

Sep. 20, 2025

Changed configuration of system to allow for faster rendering of code and diagram components.

Sep. 13, 2025

Enabled the AI Helper agent for all users.

Sep. 06, 2025

Enabled the Google Gemini-2.5 Flash Image (a.k.a. nano‑banana) image generation model for all eligible users.
Changed configuration to allow more OpenAI GPT 5 models more time to respond to complex prompts.

August 2025

Aug. 14, 2025

Enabled the OpenAI GPT 5 and GPT 5-mini models for all users.

Aug. 02, 2025

Deployed new infrastructure updates designed to resolve potential database issues. This should improve the stability of the Playground.
Fixed an issue where Google models were not creating a chat title when starting new conversations.
Fixed an issue in which the wrong name was being displayed for application shortcuts such as desktop app icons, iOS Web Clips, home screen shortcuts, and shared links.

July 2025

Jul. 30, 2025

Disabled the conversation search feature temporarily while we work to improve platform stability.

Jul. 24, 2025

AI API Gateway - Upgraded the version of LiteLLM supporting virtual API access to the models within the AI Playground.
AI API Gateway - Added additional API endpoints for models, embeddings, and generations.
AI API Gateway - Enabled the OpenAI GPT4.1 model and text embedding ada as options for virtual API access.
AI API Gateway - Enabled the Anthropic Claude 4 Sonnet model as an option for virtual API access.
AI API Gateway - Enabled the Google Gemini 2.5 Pro model and Imagen 3 model as options for virtual API access.

Jul. 16, 2025

Changed configuration of Google Gemini 2.5 Pro model to accommodate vendor change.
AI API Gateway - Changed configuration of Google Gemini 2.5 Pro model to accommodate vendor change.

Jul. 13, 2025

Changed how shared links are displayed, removing the footer links, making it easier to archive and share conversations.

Jul. 10, 2025

Updated site text for the DoResearch Policy Handbook agent.

Jul. 04, 2025

Deployed major UI enhancements and system optimizations for a better experience.
Deployed several new accessibility improvements, including:
- Enhanced accessibility and visual consistency, updating base styling and labels in multiple places.
- Added label accessible names to buttons for file attachment removal.
- Refactored the SearchBar and conversation list to remove redundant hooks for better maintainability.
- Enhanced keyboard and screen reader accessibility of the SearchBar, including appropriate labels, focused state, and tabbable logic.
- Improved Forking and SplitText accessibility design.
Transitioned existing plugins to new AI agent functionality.
Enabled the DoResearch Policy Guide and Faculty Handbook Search agents for all users.
Added a Temporary Chat feature for all users, which excludes chats from search results and automatically deletes them.
Added the option to open thinking dropdown menus for DeepSeek models.

June 2025

Jun. 26, 2025

AI API Gateway - Laid out the groundwork for enabling new models for virtual API access.
AI API Gateway - Enabled the Anthropic Claude 4 Sonnet model as an option for virtual API access.
AI API Gateway - Enabled the Google Gemini 2.5 Pro model as an option for virtual API access.

Jun. 12, 2025

Enabled the Claude 4 Sonnet model for all users.

Jun. 07, 2025

Enabled the OpenAI GPT 4.1 model for all users.
Enabled the Google Gemini 2.5 Pro model for all users.

May 2025

May 31, 2025

Fixed an issue causing frequent container restarts due to uncaught errors in the messages collection. This should improve the stability of the Playground.

May 22, 2025

AI API Gateway - Upgraded the version of LiteLLM supporting virtual API access to the models within the AI Playground.

May 15, 2025

Enabled the DeepSeek V3 model for all users.
Enabled the Meta Llama 4 model for all users.

May 01, 2025

Deployed minor UI configuration change to improve experience.

April 2025

Apr. 17, 2025

Enabled the ScholarAI plugin for all users.

Apr. 07, 2025

AI API Gateway - Enabled the AI API Gateway service to provide direct API access to models within the AI Playground.

Apr. 03, 2025

Changed allowed workgroups giving sponsored affiliates access to the AI Playground.

March 2025

Mar. 20, 2025

Enabled the Admin Guide plugin for all users.
Enabled the Google Gemini 2.0 Flash Lite model for all users.
Decommissioned Google Gemini 1.5 Flash model.
Fixed an issue with the available models in the Plugins menu.

Mar. 6, 2025

Enabled Claude 3.7 Sonnet model for all users.

February 2025

Feb. 13, 2025

Enabled the DeepSeek R1 model for all users.
Updated UI to resolve an issue resulting from deployment of the new model.
Updated site landing page text.

Feb. 8, 2025

Enabled the OpenAI GPT o1 and o3-mini models for all users.
Enabled the Google Gemini 2.0 Flash (experimental) model for all users.
Deployed improvement for security and access logging related to shared links.

Feb. 2, 2025

Fixed an issue with system memory allocation and changed the amount of dynamic memory allocated to the application.

January 2025

Jan. 29, 2025

Deployed new platform performance and stability enhancements.

November 2024

Nov. 16, 2024

Enabled the Google Gemini 1.5 Flash and 1.5 Pro for all users.
Changed how files are managed and deleted when uploaded to the Azure AI Assistants.
Changed the conversation sharing feature to require authentication to view shared links.
Fixed an issue with long text in relation to Google models that was causing unexpected errors.
Fixed an issue with user icons blocking access to part of the screen.

Nov. 2, 2024

Enabled the Imagen 3 image generation model for all eligible users.
Updated the landing page header.

October 2024

Oct. 12, 2024

Enabled the Llama 3.2 model for all users.
Deployed many accessibility updates and miscellaneous bug fixes.

Oct. 3, 2024

Added conversation bookmarks features.
Added side-by-side multi-model comparison feature.
Added app & web page rendering capabilities using Artifacts feature.

Oct. 2, 2024

Changed allowed workgroups giving students, postdocs, and visiting scholars access to the AI Playground.

September 2024

Sept. 20, 2024

Fixed an error with concurrent login attempts.
Fixed an error with max token counts.
Fixed an error with uncaught exceptions.
Fixed an issue with Anthropic models causing containers to restart.

Sept. 19, 2024

Changed allowed workgroups giving faculty and staff access to the AI Playground.
Updated site landing page text and links.
Fixed an issue with Azure Assistants causing containers to restarts.
Removed an erroneous menu option leading to unexpected errors.

Sept. 12, 2024

Enabled the OpenAI GPT 4o mini model for all users.
Changed default model which opens upon loading the webpage.
Added support link to main page.

Sept. 9, 2024

Fixed an issue with improperly copied attachments causing errors.
Fixed an error with DALL-E 3 image generation.

Sept. 7, 2024

Fixed an issue uploading documents to Anthropic models resulting in errors.
Fixed an issue causing container restarts because of authorization handling exceptions.
Fixed an error with unformatted responses.
Deployed several cosmetic changes and accessibility updates.
Migrated production UI to new cloud infrastructure host.

August 2024

Aug. 23, 2024

Enabled the Data/Code Analyst AI Assistant for all users.
Deployed several environment configuration changes.

Aug. 15, 2024

Changed allowed workgroups giving campus IT teams access to the AI Playground.
Enabled the Llama 3.1 model for all users.
Enabled the Wolfram Alpha plugin for all users.

Aug. 5, 2024

Deployed several cosmetic and accessibility updates.

Aug. 2, 2024

Enabled the DALL-E 3 image generation model for all eligible users.
Changed the authentication refresh token timeout.

July 2024

July 26, 2024

Enabled the Google Search plugin for all users.
Enabled the light weigh web page scraper plugin for all users.
Updated site landing page text.
Fixed an issue causing uncaught exception errors with database connection.