@alexandr_wang Decent model but still benchmaxxed. Not as good as v4 flash 0731, 5.3 flash or gem 3.7 flash. Used it a lot because of its low price (contributor tier) but it’s lacking intelligence and hallucinates a bit too much.
@ClementPillette@kevinkern I don’t know why High isn’t the default thinking effort. It has nearly the same capabilities while using half as many tokens.
@1squadramilano@leploutos@theo_louro Abuse pas c’est un très bon modèle pour la taille mais niveau finesse d’analyse et connaissances générales on est encore loin d’Opus 4.6