I-OpenAI isheshisa i-GPT-5.2 ukuze iphendule ekuphusheni kwe-Google Gemini 3

Umjaho wobuholi ku ubuhlakani bokwenziwa kokukhiqiza Ukushuba kwesimo kunyukile emasontweni adlule. Kulandela umthelela we-Gemini 3, imodeli entsha ye-Google, I-OpenAI inyathele igesi futhi ilungiselela ukukhishwa kwangaphambi kwesikhathi kwe-GPT-5.2, ukuthuthukiswa okubalulekile kubuchwepheshe bayo obufuna ukuvala, ngisho nokuhlehlisa, igebe lokusebenza elivulwe isimbangi sayo esikhulu.

Imithombo esondelene nezinhlelo zangaphakathi ze Inkampani ikhombisa ukuthi inguqulo entsha yemodeli yokucabanga iyi Ilungele ukungena ekukhiqizeninokuthi abaphathi banikeze umyalelo wokubeka phambili ukuthunyelwa kwawo ngaphezu kwamanye amaphrojekthi. Lokhu akukhona ukulungisa okuncane okulula: I-GPT-5.2 isilungele ukuba yimpendulo yokuqala enkulu ye-OpenAI ekuthuthukisweni kwe-Google kumabhentshimakhi, ukuhlolwa kokucabanga, kanye nombono womphakathi.

"Ikhodi ebomvu" ethuthukisa ishejuli ye-GPT-5.2

Ngokwemibiko ehlukene yangaphakathi eputshuzelwe emithombo yezindaba eyisipesheli, Sam Altman, i-CEO ye-OpenAI, Kubikwa ukuthi inkampani imemezele isimo "esibomvu" ngemuva kokubukeza ukusebenza kweGemini 3 emasontweni asanda kwedlula.Lokhu kuthutha kuhilela ukuqondisa kabusha izinsiza ngokushesha kumongo we-ChatGPT kanye nokusheshisa ukuthuthukiswa kwemodeli enkulu elandelayo.

Izinhlelo zakuqala zabeka ukufika kwe GPT-5.2 kuya phambili Ekupheleni kukaDisembaKulandelwa ijubane elibekwe iziphindaphindo zangaphambilini: I-GPT-5 kanye ne-GPT-5.1, esetshenziswe ezinyangeni ezintathu kamuva. Nokho, Gemini 3 ukusebenza emisebenzini yokucabanga ethuthukisiwe, amamethrikhi aseduze ne-AGI namandla e-multimodal iphoqe i-OpenAI ukuthi yenze umnyakazo ngaphambi kwesikhathi.

Manje, inkampani isebenza ngedethi ezungezwe ngokubomvu ekhalendeni layo langaphakathi: the ULwesibili, Disemba 9Imithombo ehlukahlukene iyavuma ukuthi lolu usuku okuhloswe ngalo ukubeka i-GPT-5.2 ezandleni zomphakathi, nakuba umnyango ushiywe uvulelwe ukukhishwa okunezigaba noma ukubambezeleka okuncane kuye ngomthamo weseva kanye nokulungiswa kwakamuva kobuchwepheshe.

Umlayezo ovela kubaphathi ukuthi ukubuyekezwa kuyaqhubeka. "ikhiyiwe yalayishwa"Ilungele ukuqalisa uma kungekho zihibe ezinkulu eziphakamayo. Noma kunjalo, endaweni eyinkimbinkimbi njengaleyo yezinhlobo zezilimi ezinkulu, izinsuku zangaphakathi zibhekwa njengeziqondiso esikhundleni sezithembiso eziqinile.

Kulo mongo, isinqumo sika-Altman akuyona nje indaba yokuhlonishwa kwezobuchwepheshe: futhi sihlose ukuqinisekisa abatshalizimali, abalingani kanye namakhasimende amakhulu abaqaphela eduze noma yikuphi umnyakazo we-Google, i-Anthropic, noma i-Meta phezulu emakethe ye-AI.

I-Gemini 3 ibeka ingcindezi futhi iphoqa i-OpenAI ukuthi isabele

I-trigger yalolu shintsho lwesu kube ukuvela kwe IGemini 3, imodeli ye-Google okuthe emasontweni ambalwa nje igxumele phezulu ezilinganisweni zokusebenzaUkuhlola okuhlukile okuzimele kuyibeka ngaphambi kwamamodeli angcono kakhulu asesidlangalaleni e-OpenAI emisebenzini ye ukucabanga okuthuthukile kanye nezinkomba ezimbalwa zesitayela se-AGI.

Le modeli entsha ayizange ivelele kuphela ngamakhono ayo e-multimodal, kodwa futhi ukusebenza okuqinile ekuhlolweni okuyinkimbinkimbiLokhu kudale ukusabela kokumangala ngaphakathi komkhakha wezobuchwepheshe. Eminye imithombo ibonisa ukuthi ukuqhathanisa kwangaphakathi kwe I-OpenAI ngabe ikubeke kwacaca lokho, ezimeni ezithile, I-Gemini 3 yayihamba phambili ngemajini ekhathazayo kumamodeli wendlu.

Ibhekene nalesi simo, inkampani ka-Altman ikhethe impendulo eqondile: sheshisa i-GPT-5.2 njengempendulo yokuqala enamandla ekuthuthukisweni kwe-GoogleImibiko eminingana iveza ukuthi u-Altman ngokwakhe wavuma ngasese ukuthi igebe elivulwe yi-Gemini 3 lidinga ukuvalwa “ngokushesha okukhulu,” kukho kokubili isithombe kanye nokuncintisana kwangempela kwansuku zonke.

Ngaleso sikhathi, i-Anthropic isaqhubeka nokuqhubekela phambili nomndeni wayo wakwaClaude, kanti iMeta iqinisa isu layo eliyimodeli evulekile, inikele endaweni lapho inyanga ngayinye ibala Futhi noma yikuphi ukubambezeleka kungahumusheka ekulahlekelweni yisizinda kwezinkampani, onjiniyela, nokuphathwa komphakathi eYurophu nasemhlabeni wonke.

Emakethe yaseYurophu, lapho zombili izikhungo nezinkampani zifuna izixazululo ze-AI enokwethenjelwa, efundekayo futhi eqinisekisiweUmbono wokuthi ubani onokuhola kwezobuchwepheshe ubaluleke kakhulu lapho ukhetha umhlinzeki wamaphrojekthi wokuzenzakalela, ukuhlaziya idatha, noma abasizi ababonakalayo.

Okuthenjiswa yi-GPT-5.2: iziqhumane ezincane kanye nemisipha eyengeziwe

Kude nokugxila ezicini eziwubukhazikhazi noma ukuhlola okuhlala isikhathi esifushane, i-OpenAI iqondise i-GPT-5.2 ochungechungeni lwentuthuko ecaciswe kakhulu. Imibiko yangaphakathi kanye nokuputshuka kwabezindaba kuyavuma ukuthi inguqulo entsha izogxila kukho isivinini, ikhwalithi yokucabanga, nokwethembekaLezi zici ezintathu, ekusebenzeni, zenza umehluko kulabo abasebenzisa i-ChatGPT nsuku zonke.

Inkampani ifuna ukunikeza imodeli izimpendulo ezisheshayoikakhulukazi ekubonisaneni okude noma okuyinkimbinkimbi, ukunciphisa izikhathi zokulinda abanye abasebenzisi abaziqaphele ezinyangeni zakamuva. Ngesikhathi esifanayo, umsebenzi uyaqhubeka ku-a ukucabanga okucwebe kakhudlwana, kunokungahambisani okunengqondo okumbalwa kanye nezinga lephutha eliphansi emisebenzini yobuchwepheshe noma yezinyathelo eziningi.

Okunye okubalulekile yi- ukunciphisa ukwehluleka kanye nezimpendulo ezingaphandle kokuqukethwe. I-GPT-5.2 kufanele yenze okunganembile okumbalwa, igcine intambo ibe ngcono ezingxoxweni ezinde, futhi isingathe imiyalelo enemininingwane ngokuqinile—ukhiye wokusetshenziswa kochwepheshe ezinkampanini zase-Europe, ezokuphatha, nezikhungo zocwaningo.

Kukhona futhi ukugxila okukhethekile ku ukuqonda umongo futhi i ukuzivumelanisa nezimo yemodeli, okungukuthi, ikhono layo lokulungisa ithoni, ileveli yemininingwane, noma izidingo ezithile zomsebenzisi ngaphandle kokulahlekelwa ukunemba. Lokhu kuthuthukiswa kubaluleke kakhulu ezinhlelweni zokusebenza emikhakheni efana nokubonisana, ukumaketha, imfundo, noma ukuthuthukiswa kwesofthiwe.

Emkhakheni we-multimodal, nakuba inguquko ephelele ingalindelekile, inhloso iwukuba Lungiselela indlela imodeli ehlanganisa ngayo umbhalo nezinye izinhlobo zokufakwayo, eqinisa ukusetshenziswa okusebenzayo ku imisebenzi ye-multimodal lapho amadokhumenti, izithombe, noma idatha ehlelekile ixutshwa khona.

Shift kokubalulekile: ukusebenza ngaphezu kwezici ezintsha eziwubukhazikhazi

Ukukhishwa kwe-GPT-5.2 kuza nokushintshwa kwamasu ngaphakathi kwe-OpenAI. Ngokwemithombo ekhonjiwe, umyalelo kaSam Altman ucacile: ukubeka imisebenzi emangalisayo ngemuva futhi ugxile kulokho abasebenzisi bamandla namakhasimende ezinkampani abazisa kakhulu.

Lokhu kuhumushela ekumiseni isikhashana noma ekunciphiseni izinhlelo ezifana nalezi izinhlelo ezintsha zokwenza imali ku-ChatGPT, ukuhlanganiswa kokukhangisa, noma ukuthunyelwa kwe ama-agent wesikhathi eside we-AI Lawa maphrojekthi, agxile ezindaweni ezibucayi njengezempilo noma ukuthenga komuntu siqu, asekhona etafuleni, kodwa athathe isihlalo esingemuva.

Okuhamba phambili kwamanje ukuqinisa umnyombo wesevisi: ukuzizwisa okuzinzile okunokuhamba kancane, ukuphatha okuqinile kwemithwalo esindayo, nomuzwa wokuthi imodeli "icabanga kangcono" futhi yenza amaphutha ambalwa. Ngokwamagama abanye abahlaziyi, i-OpenAI ikhethe "Umsindo omncane kanye nemisipha eyengeziwe".

Le filosofi ihambisana nezidingo zenani elikhulayo lezinkampani zaseYurophu ezihlanganisa i-ChatGPT ezinqubweni zazo zangaphakathi: kusukela kumajeneretha asalungiswa nabasizi abasekelayo kuye kumathuluzi okuhlaziya imibhalo yezomthetho, yezobuchwepheshe, noma yezezimali. Kulezi zimo zokusetshenziswa, i- ukwethembeka okuqhubekayo ngokuhamba kwesikhathi Ibiza ngaphezu kwanoma yikuphi ukusebenza okukodwa kombukiso.

Ezingeni lomkhiqizo, ukuthuthukiswa okuningi ku-GPT-5.2 kulindeleke ukuthi kubonakale njenge amashifu athuleNgeke kube khona ukuklanywa kabusha kwesixhumi esibonakalayo esikhulu noma "ubuso" be-chatbot entsha, kodwa kunalokho ukulungiswa kwangaphakathi okubonakala nsuku zonke, indlela esabela ngayo, kuthatha isikhathi esingakanani, nokuthi igcina kangakanani ukuvumelana.

Umjikelezo wokubuyekeza okhula ngokushesha

Ukwethulwa kwe-GPT-5 ehlobo, kulandelwe yi-GPT-5.1 ngoNovemba, futhi manje sekuseduze ukufika kwe I-GPT-5.2 ngaphansi kwenyangaBadweba iphethini ecacile: I-OpenAI ingene emjikelezweni wokuphindaphinda okusheshayo, omiswe kokubili ukuthuthuka kwayo kanye nezinyathelo ezinonya zezimbangi ezifana ne-Google ne-Anthropic.

Kulo mongo, izinsuku zokukhishwa ziya ngokuya ziba kushintshashintshaIzinhlelo zingase zishintshe ngomzuzu wokugcina ngenxa yezinkinga zengqalasizinda, ukulungiswa kwezokuphepha, ukuhlolwa komthwalo, noma izinqumo zamasu eziphendula izimemezelo zezimbangi. Noma kunjalo, ukuphikelela kwangomhla ziyisi-9 kuZibandlela njengokuhloswe kwangaphakathi kuphakamisa ukuthi ukuthuthukiswa kuyaqhubeka.

Emphakathini wabasebenzisi nabathuthukisi base-Spain nakulo lonke elaseYurophu, lokhu ushintsho oluncane Inomkhawulo ophindwe kabili: ngakolunye uhlangothi, iqinisekisa ukuthi ithuluzi lihlala lithuthuka; ngakolunye, iphoqelela ukuzivumelanisa nendawo lapho amakhono emodeli angashintsha ngokuphawulekayo endabeni yamaviki.

Izinkampani, izitudiyo zokuthuthukisa, nezinhlaka zikahulumeni eseziqale izinsiza zokwakha ku-OpenAI API kufanele zibhekisise ukuthi kanjani I-GPT-5.2 ishintsha ukuziphatha kwemodeliikakhulukazi emisebenzini enethonya elikhulu njengokwenza ikhodi, usizo lwezomthetho, noma ukuhlaziywa kwedatha okuyimfihlo.

Ngesikhathi esifanayo, ingcindezi yokuncintisana ikhuthaza bonke abadlali emkhakheni—OpenAI, Google, Anthropic, Meta—ukuqinisa umsebenzi wabo ukuphepha, ukunciphisa ukwenzelela, nokulawula ukubona izinto ezingekho, izici umlawuli waseYurophu azinaka kakhulu ngaphakathi kohlaka lomthetho omusha we-AI.

Umthelela kubasebenzisi nakumabhizinisi aseSpain naseYurophu

Nakuba i-OpenAI ingakanikezi imininingwane yokuthi ukukhishwa kwesifunda kuzohlelwa kanjani, kuvamile ukuthi izici ezintsha eziyinhloko ze-ChatGPT kanye ne-API yayo ifike cishe kanye kanye I-United States naseYurophu, ngomehluko omncane kuye ngohlobo lwesevisi nezibopho zokulawula zendawo ngayinye.

Kubasebenzisi ngabanye, Ukuthuthukiswa okuphawuleka kakhulu ku-GPT-5.2 kuzoba umuzwa wokushesha okukhulu kanye nokuthambekela okuncishisiwe kwamaphutha. ezincazelweni eziyinkimbinkimbi, izifinyezo ezinde, noma imisebenzi yokudala enezinyathelo eziningi. Ngamanye amazwi, i-ChatGPT ebonakala "ibambeka" kancane futhi igcina ukugeleza okungcono uma izingxoxo ziba zinde.

Emkhakheni webhizinisi, ikakhulukazi ezindaweni ezifana namabhange, amandla, ezokuxhumana, kanye nokubonisana nokuba khona okuqinile eSpain nase-EU, ukuzibophezela kwe-OpenAI isivinini, ukuzinza kanye nokwenza ngokwezifiso block nge isidingo sabasizi bangaphakathi abakwazi ukuphatha imiqulu emikhulu yolwazi ngokugadwa okuncane.

Ukucabanga okuthuthukisiwe nakho kuyisihluthulelo samaqembu we ukuthuthukiswa kwesofthiwe nokuhlaziywa kwedathaLawa mamodeli asetshenziselwa ukubuyekeza ikhodi, ukuhlola imibono, noma ukukhiqiza imibhalo yezobuchwepheshe. Izinga lokwehluleka eliphansi kanye nokucabanga okuqinile kungahumushela ku... ukonga isikhathi kanye nezibuyekezo ezimbalwa ezilandelayo.

Ngakolunye uhlangothi, isinqumo sokuthi misa okwesikhashana ukuthunyelwa kwama-ejenti e-AI azimele, ikakhulukazi ezindaweni ezibucayi njengezempilo, Lokhu kuhambisana nesexwayiso izikhungo eziningi zaseYurophu ebezisifuna.Ngaphambi kokugxumela ezinhlelweni ezenza izinqumo ezizimele, iBrussels kanye nabalawuli kazwelonke bafuna iziqinisekiso eziqinile zokubeka izinto obala, ukwengamela komuntu, kanye nokulandeleka.

Sekukonke, i-GPT-5.2 ibumba isibuyekezo esidizayinelwe qinisa ukwethembana ekusebenziseni imodeli yobungcweti nansuku zonke, isici esibaluleke kakhulu ezimakethe lapho ukulawulwa kwe-AI kuthuthuka ngokushesha futhi amaphrojekthi amakhulu ahlaziywa ngekama lamazinyo amahle.

Umbango obeka ijubane lokuqamba izinto ezintsha

Isinqumo se-OpenAI sokuvula "ikhodi ebomvu" nokuthuthukela phambili kwe-GPT-5.2 sibonisa izinga ukuncintisana ne-Google Isetha ijubane lokuqamba izinto ezintsha ku-AI namuhla. Uma eminyakeni embalwa edlule kwaba ukuvela kwe-ChatGPT okwaphoqa i-Google ukuthi isabele ngokushesha, manje sekuyi-Gemini 3 ephoqa i-OpenAI ukuthi ihlele kabusha izinto eziza kuqala.

Lokhu okuguquguqukayo akukhawulelwe ekukhangiseni: imodeli ngayinye entsha ephakamisa izinga lokusebenza iphoqa bonke abanye ukuthi babuyekeze imephu yemigwaqo yabo, banikeze kabusha amaqembu, futhi benze izinqumo ezinzima mayelana nokuthi imaphi amaphrojekthi ahlehlisiwe futhi yiziphi eziba ngamasu. Kulokhu, umlayezo wangaphakathi kwa-OpenAI yilokho ukuqina kwemodeli yesisekelo Inesisindo esingaphezu kwanoma yikuphi ukusebenza okwengeziwe.

Ku-ecosystem yobuchwepheshe baseYurophu, lo mzabalazo unomkhakha owengeziwe. Ukuhlanganiswa kwamamodeli amakhulu ambalwa okulinganisa, alawulwa yizinkampani zase-US, kuhambisana nemizamo ye-European Union yokukhuthaza ezinye izindlela zayo kanye nokuqinisekisa ukuthi ukuthunyelwa kwe-AI kuhambisana nemithetho yayo. ubumfihlo, ukuphepha kanye namalungelo ayisisekelo.

Ngaleso sikhathi, kokubili ama-startups amancane nezinkampani ezinkulu kulo lonke izwekazi Balandela eduze konke umnyakazo kusuka ku-OpenAI ne-Google, ngokwazi ukuthi i amakhono kanye nemikhawulo yamamodeli abo Azothinta amaphrojekthi e-automation, abasizi be-virtual, izibalo ezithuthukile, kanye nenkonzo yamakhasimende ezinyangeni ezizayo.

Kulesi simo esishintsha ngokushesha, i Ukukhishwa kwangaphambi kwesikhathi kwe-GPT-5.2 kuhunyushwa njengomzamo wokuthola enye yezindawo ezingokomfanekiso nezobuchwepheshe ezisanda kulahleka ku-Gemini 3.ukuqinisa umqondo wokuthi impi yobuholi obukhiqizayo be-AI manje isiliwa isonto nesonto, hhayi ngemijikelezo yonyaka.

Njengoba umhla ziyisi-9 kuZibandlela usethwe njengosuku okuhloswe ngalo kanye nemephu yomgwaqo egxile ejubaneni, ekucabangeni nasekuthembekeni, I-GPT-5.2 ihlose ukubuyisela i-ChatGPT lowo muzwa wemodeli eqinile neshesha kakhulu abasebenzisi abaningi abebengekho.Ngenkathi ithumela umlayezo ocacile emqhudelwaneni: I-OpenAI ayizimisele ukuvumela i-Google ukuthi ihlelele isizukulwane esilandelayo sobuhlakani bokwenziwa.

I-athikili ehlobene:

Isetshenziswa kanjani i-Meta's MusicGen endaweni ngaphandle kokulayisha amafayela efwini

U-Alberto Navarro

Ngingumshisekeli wezobuchwepheshe oguqule izintshisekelo zakhe "ze-geek" zaba umsebenzi. Ngichithe iminyaka engaphezu kwengu-10 yempilo yami ngisebenzisa ubuchwepheshe obusezingeni eliphezulu kanye nokukitaza ngazo zonke izinhlobo zezinhlelo ngenxa yelukuluku lokufuna ukwazi. Manje sengiqeqeshelwe ubuchwepheshe be-computer nemidlalo yama-video. Lokhu kungenxa yokuthi sekuphele iminyaka engaphezu kwengu-5 ngisebenza ngokubhalela amawebhusayithi ahlukahlukene ezobuchwepheshe nemidlalo yevidiyo, ngenza izindatshana ezifuna ukukunikeza imininingwane oyidingayo ngolimi oluqondakala yiwo wonke umuntu.

Uma unemibuzo, ulwazi lwami lusukela kuyo yonke into ehlobene nesistimu yokusebenza ye-Windows kanye ne-Android yomakhalekhukhwini. Futhi ukuzibophezela kwami ​​​​kuwe, ngihlala ngizimisele ukuchitha amaminithi ambalwa futhi ngikusize uxazulule noma yimiphi imibuzo ongase ube nayo kulo mhlaba we-inthanethi.