AI COURSE WORKBENCH

Learning Pack AI

Choose a faculty and class. AwardSpace hosts the website and PHP backend. The backend calls your Google VM language and vision models and repairs the LaTeX. PDF compilation is delegated to the Google VM, so AwardSpace does not need a local TeX installation.

gpt-oss + optional local vision · real PDF
2Configure the pack
No class selected yet.
Enter the public model endpoint as seen by AwardSpace PHP. Recommended: an HTTPS reverse-proxy URL on your Google VM. This browser remembers your last address.
Pack contents
3Optional source material

Upload a PDF to make it the primary source, or paste notes. AwardSpace PHP extracts the source text/pages when the required server tools are available. Optional vision mode sends selected page images to your Google VM vision model and lets gpt-oss place useful source images in the final PDF.

The vision address is always shown. Enable the checkbox above only when you want an uploaded textbook PDF analyzed. Recommended endpoint: https://api.maximinfo.com/vision/v1/chat/completions.
4Generate LaTeX + compiled PDF

Generation pipeline: When vision mode is off, AwardSpace PHP makes one gpt-oss writing request. When vision mode is on, Qwen3-VL first analyzes a few relevant PDF pages, then gpt-oss writes the LaTeX workbook. AwardSpace sends the finished LaTeX to the Google VM PDF renderer, which compiles the real PDF with LuaLaTeX.