At the start of the new semester, RWTHgpt receives a comprehensive update. This provides extended functionalities, improved usage options, and most notably, access to new, powerful models.
Newly available are two additional language models OpenAI o1 and OpenAI o3-mini. In addition, RWTHgpt now supports models hosted directly by OpenAI.
Token limits have been significantly increased, especially for students: instead of weekly, there are now daily token quotas. For gpt-4o-mini, 25,000 tokens are now available per day, and for gpt-4o, 6,000 tokens per day – previously this was the weekly limit. The newly introduced models OpenAI o3-mini and o3-mini-high each offer 12,000 tokens daily. The new model OpenAI o1 can be used with up to five prompts per day.
The update also includes functional enhancements, such as the implementation of the chat search in the sidebar and an optimized display of the prompt input field.
On the backend side, multi-tenancy has been introduced, prompt processing has been improved, and display errors have been fixed, such as the correct display of user initials and the correct output of the “&” character in generated code. Preparations have also been made for the integration of open-source models.
Instructions for use can be found on IT Center Help.