Builds are faster.
Typical LaTeX projects now compile in ~1/3 the time on our platform. Incremental caching, smarter engine selection, and parallel runs mean fewer timeouts and faster feedback for users.
Same requests, same files - just quicker PDFs.