Hacker News new | ask | show | jobs
by ei625 497 days ago
For 1, they apply non-reasoning model then apply reasoning model.