OpenAI waxay horumarisay sirdoon macmal ah oo leh moodelkeeda cusub, OpenAI-o1, oo maanta la shaaciyay, kaas oo si macquul ah u xallin kara dhibaatooyinka adag iyada oo aan loo baahnayn in la ballaariyo sida kii ka horreeyay, GPT-4. Marka laga reebo moodooyinka luqadda weyn ee caadiga ah (LLMs) oo bixiya jawaabo hal tallaabo gudahood, OpenAI-o1 wuxuu ku mashquulsan yahay xallinta talaabo talaabo, isagoo go'aaminaya jawaabaha si aadanaha u eg. Mira Murati, oo ah sarkaalka tignoolajiyada sare ee OpenAI, waxay carrabka ku adkeysay isbeddelkan habka, iyadoo sheegtay in OpenAI-o1 uu ku fiican yahay xallinta hawlaha sababta adag ee moodooyinka hadda jira, oo ay ku jiraan GPT-4o, ku dhibtooday. Inkasta oo OpenAI ay horumarinayso GPT-5 si uu uga weyn yahay GPT-4o, waxay qorsheynayaan inay ku daraan awoodaha sababidda ee lagu soo bandhigay OpenAI-o1. Modelkan cusub wuxuu isticmaalaa barashada xoojinta si uu u xoojiyo awoodda sababtiisa, isagoo bixinaya jawaab celin ku saabsan jawaabihiisa si uu uga fiicnaado habkiisa. Mark Chen, Madaxa cilmi-baarista ee OpenAI, ayaa soo bandhigay awoodda moodelkan iyadoo la xallinayo dhibaatooyinka kiimikada heerka sare iyo dhibaatooyinka xisaabta ee adag ee moodooyinka hore aan xallin karin. OpenAI-o1 ayaa soo bandhigay horumar weyn oo ka sarreeya GPT-4o, gaar ahaan maadooyin kala duwan, isagoo gaaray 83% saxnimada Imtixaanka Xisaabta ee Martiqaadka Mareykanka, marka la barbardhigo 12% ee GPT-4o.

Si kastaba ha ahaatee, OpenAI-o1 waa ka gaabis badan yahay oo ma laha astaamo sida raadinta webka iyo awoodaha kala duwan ee modal-ka, sida turjumaada sawiro ama cod. Dadaalka lagu horumarinayo sababta LLMs waa mawduuc caadi ah oo lagu sameeyo cilmi-baarista AI, iyadoo tartamo sida Google ay raacayaan waddooyin la mid ah. Habka OpenAI waxay mataleysaa booddo ballaaran oo ku wajahan sababta guud, halka khubaradu ay carrabka ku adkeeyaan muhiimadda ah in la fahmo sida moodelladani u soo saaraan xalalkooda, iyadoo la tixgelinayo saamaynta ay ku yeelan karaan go'aan qaadashada. Murati waxay xustay in OpenAI-o1 sidoo kale muujinayo is-waafajin wanaagsan oo la jaanqaadaya heerarka anshaxa, iyadoo ay sababtay fikrada ficilkeeda, iyadoo yareyneysa waxyeelada ka soo bixi karta. Khabiirka AI Oren Etzioni wuxuu carrabka ku adkeeyay baahida loo qabo LLMs inay ku hawlgalaan xallinta dhibaatooyinka talaabo talaabo ah, iyadoo kaliya kor u qaadista aysan ku filnayn horumarka. Chen wuxuu soo gabagabeeyay in habkan sababta cusub uu awood u siin karo horumarinta AI ee hufan iyo kharash-ku-ool ah, isagoo la jaanqaadaya yoolka aasaasiga ah ee OpenAI.