>>717790462
various llms do in fact solve competition programming challenges pretty easily, openai, claude, etc. have all demonstrated this capability in the public sphere, google has various models which do really well, etc.
this is like 4+ years old news at this point, to deny it is just showing you don't pay attention to the field
now do these companies exaggerate how well their internal, nonpublic, nonverifiable models are? lmao of course, there's massive financial incentive to do so
but the ones out in the open you can literally either locally host or prompt yourself to show that they can