Now O1 istheirquote-unquotereasoningmodel, anditscoresreallyhighoncodingandmathbenchmarks.
Itfeelssimilartoothermodelsbuthastheabilitytopromptitselfoverandoveragainusing a techniquecalledChainofThoughtwhichgeneratesreasoningtokens, andhastheabilitytosolvemorecomplexproblems, especiallywhenitcomestomathandprogramming.
However, thisrequires a lotmorecomputingpower, andontheproplan, yougetsomethingcalledProModetoapparentlydoevenbetteronthesebenchmarks.
AndasanearlyadopterofanyAIsloththey'llfeedme, I canconfirmthat O1 ProModeisactuallyquiteimpressive.
Italmostcorrectlybuilt a Svelte 5 app, andjustlookatthis $200 SVGunicorn.
That's impressive, but I stilldon't think I'llbebuyingitformykidsthisFestivus.
Overthelastyear, Altmanhasbeentryingtousethegovernmenttoachieveregulatorycaptureandcrushthiscompetition, butnowit's toolate, andhe's facing a bigElonproblem.
Yousee, a fewyearsago, ElonMuskoverpaidfor a failingsocialmediawebsitecalledTwitter, sohecouldbanthiscollegekid's accountwhomade a Pythonapptotrackthelocationofhisprivatejet.
Everyonelaughedathim, butitturnsoutitwasinfact a 5D chessmove.
HeparlayedhisTwitterpurchasetomakeanothercrazybettosupportTrumpforpresident, andnowthatbetispayingoffin a bigway, becauseguesswhojustbecameTrump's AIczar?