I-OpenAI ikhawuleza i-GPT-5.2 ukuphendula kwi-Google Gemini 3's push

Ugqatso lobunkokeli kwi ubukrelekrele obuvelisayo Uxinzelelo lonyukile kwezi veki zidlulileyo. Ukulandela impembelelo yeGemini 3, imodeli entsha kaGoogle, I-OpenAI iye yanyathela igesi kwaye ilungiselela ukukhutshwa kwangaphambili kwe-GPT-5.2, uphuculo oluphambili kwitekhnoloji yalo efuna ukuvala, kunye nokuba umva, umsantsa wokusebenza ovulwe ngulowo ukhuphisana naye ophambili.

Imithombo ekufutshane nezicwangciso zangaphakathi ze Inkampani ibonisa ukuba inguqulelo entsha yemodeli yokuqiqa Ukulungele ukuya kwimvelisokunye nokuba abaphathi banike umyalelo wokubeka phambili ukusasazwa kwabo ngaphezu kwezinye iiprojekthi. Olu asilohlengahlengiso oluncinci olulula: I-GPT-5.2 ilungele ukuba yimpendulo yokuqala ye-OpenAI enkulu kwinkqubela kaGoogle kwiibenchmarks, iimvavanyo zokuqiqa, kunye nembono yoluntu.

I-"code red" ephucula ishedyuli ye-GPT-5.2

Ngokweengxelo ezahlukeneyo zangaphakathi ezikhutshwe kumaziko eendaba akhethekileyo, USam Altman, i-CEO ye-OpenAI, Inkampani kuthiwa ibhengeze imeko "yekhowudi ebomvu" emva kokuphonononga ukusebenza kweGemini 3 kwiiveki zamva nje.Eli nyathelo libandakanya ukuhanjiswa kwezixhobo ngokungxamisekileyo ukuya kumbindi we-ChatGPT kunye nokukhawulezisa imodeli elandelayo yophuculo.

Izicwangciso zokuqala zabeka ukufika kwe I-GPT-5.2 ukuya phambili Ekupheleni kukaDisembaUkulandela isantya esisetwe ngophindaphindo lwangaphambili: GPT-5 kwaye emva koko i-GPT-5.1, isetyenziswe kwiinyanga nje ezintathu kamva. Nangona kunjalo, Gemini 3 ukusebenza kwimisebenzi yokuqiqa ephucukileyo, iimetrics ezikufutshane ne-AGI kunye nezakhono ezininzi inyanzelise i-OpenAI ukuba ithathe inyathelo phambi kweshedyuli.

Ngoku, inkampani isebenza kunye nomhla ojikelezwe ngombala obomvu kwikhalenda yayo yangaphakathi: i NgoLwesibini, nge-9 kaDisembaImithombo eyahlukeneyo iyavuma ukuba lo ngumhla ojoliswe kuyo ukubeka i-GPT-5.2 ezandleni zoluntu, nangona umnyango ushiywe uvulekile ukukhutshwa kwesigaba okanye ukulibaziseka okuncinci kuxhomekeke kumthamo weseva kunye nohlengahlengiso lwamva nje lobugcisa.

Umyalezo ovela kubaphathi kukuba uhlaziyo luyaqhubeka. "itshixiwe yalayishwa"Ilungele ukuqaliswa ukuba akukho ziphazamiso zinzima zivelayo. Nakuba kunjalo, kwimeko-bume entsonkothileyo njengaleyo yemizekelo yolwimi olukhulu, imihla yangaphakathi igqalwa njengesikhokelo kunezithembiso eziqinileyo.

Kulo mxholo, isigqibo sika-Altman ayisiyonto nje yodumo lwetekhnoloji: sikwajolise kuyo. ukuqinisekisa abatyali-zimali, amaqabane kunye nabaxhasi abakhulu abajonga ngokusondeleyo nayiphi na intshukumo kaGoogle, iAnthropic, okanye iMeta phezulu kwintengiso ye-AI.

I-Gemini 3 ibeka uxinzelelo kwaye inyanzela i-OpenAI ukuba isabele

Isiqalo solu tshintsho lweqhinga ibe kukuvela kwe IGemini 3, imodeli kaGoogle ethe kwiiveki nje ezimbalwa yatsibela phezulu kumanqanaba entsebenzoUvavanyo olwahlukeneyo oluzimeleyo luyibeka phambili kwezona modeli zibalaseleyo zikawonke-wonke ze-OpenAI kwimisebenzi ye ukuqiqa okuphezulu kunye nezalathi ezininzi ze-AGI-style.

Le modeli intsha ayizange igqame kuphela kubuchule bayo be-multimodal, kodwa nakwi ukusebenza ngamandla ngakumbi kwiimvavanyo ezinzimaOku kubangele ukusabela okumangalisayo kwicandelo letekhnoloji. Eminye imithombo ibonisa ukuba uthelekiso lwangaphakathi lwe I-OpenAI yayiza kuyenza yacaca loo nto, kwiimeko ezithile, IGemini 3 yayitsala phambili ngomda oxhalabileyo kwiimodeli zendlu.

Ukujongana nale meko, inkampani ka-Altman ikhethe impendulo ethe ngqo: isantya se-GPT-5.2 njengempendulo yokuqala enamandla kuGoogle kwangaphambiliIingxelo ezininzi zichaza ukuba u-Altman ngokwakhe wavuma ngasese ukuba isithuba esivulwe yiGemini 3 kufuneka sivalwe "ngokukhawuleza," zombini umfanekiso kunye nokhuphiswano lwemihla ngemihla.

Ngeli xesha, i-Anthropic iyaqhubeka nokutyhalela phambili nosapho lwayo lukaClaude, kwaye iMeta iqinisa imodeli yayo evulekileyo, inegalelo kwimeko apho inyanga nenyanga zibaliwe Kwaye nakuphi na ukulibaziseka kunokuguqulela ekubeni yilahleko yomhlaba kwiinkampani, abaphuhlisi, kunye nolawulo loluntu eYurophu nakwihlabathi liphela.

Kwimarike yaseYurophu, apho zombini amaziko kunye neenkampani zikhangela izisombululo ze-AI ithembekile, iyaphicothwa kwaye iqinisekisiweUmbono wokuba ngubani okhokelayo wezobuchwepheshe kubaluleke kakhulu xa ukhetha umboneleli kwiiprojekthi ezizenzekelayo, uhlalutyo lwedatha, okanye abancedisi abangabonakaliyo.

Yintoni iGPT-5.2 ithembisayo: iifireworks ezincinci kunye nezihlunu ezininzi

Kude nokujolisa kwiimpawu ezikhanyayo okanye iimfuniselo zexesha elifutshane, i-OpenAI iqhube i-GPT-5.2 kuluhlu lophuculo oluthile kakhulu. Iingxelo zangaphakathi kunye nokuvuza kumajelo eendaba kuyavuma ukuba inguqulelo entsha iya kugxila kuyo isantya, umgangatho wokuqiqa, nokuthembekaLe miba mithathu, ekusebenzeni, yenza umahluko kwabo basebenzisa i-ChatGPT yonke imihla.

Inkampani ifuna ukubonelela ngemodeli iimpendulo ezikhawulezayongakumbi kwiindibano ezinde okanye ezinzima, ukunciphisa amaxesha okulinda athe abanye abasebenzisi bawaqaphela kwiinyanga zamva nje. Kwangaxeshanye, umsebenzi uyaqhubeka kwi a ukuqiqa okusulungekileyo, kunye nokungahambelani kwengqiqo okumbalwa kunye nesantya esisezantsi sempazamo kwimisebenzi yobugcisa okanye yamanyathelo amaninzi.

Enye into ephambili yi ukunciphisa ukusilela kunye neempendulo ezingaphandle komxholo. I-GPT-5.2 kufuneka yenze ukungachaneki okumbalwa, igcine intambo ingcono kwiingxoxo ezinde, kwaye iphathe imiyalelo eneenkcukacha ngakumbi ngokuqinileyo-isitshixo sokusetyenziswa kweengcali kwiinkampani zaseYurophu, ulawulo kunye namaziko ophando.

Kukho kwakhona ugxininiso olukhethekileyo kwi ukuqonda umxholo kunye ukuba bhetyebhetye yemodeli, oko kukuthi, ukukwazi ukulungelelanisa ithoni, inqanaba leenkcukacha, okanye iimfuno ezithile zomsebenzisi ngaphandle kokulahlekelwa ukuchaneka. Olu phuculo lufanelekile ngokukodwa kwizicelo kumacandelo afana nokubonisana, ukuthengisa, imfundo, okanye uphuhliso lwesoftware.

Kwintsimi ye-multimodal, nangona i-revolution epheleleyo ayilindelekanga, injongo kukuba Lungiselela indlela imodeli edibanisa ngayo isicatshulwa kunye nezinye iintlobo zegalelo, ukuqinisa ukusetyenziswa okusebenzayo kwi multimodal workflows apho amaxwebhu, imifanekiso, okanye idatha eyakhiweyo ixutywe.

Shift kwizinto eziphambili: ukusebenza ngaphezulu kweempawu ezintsha ezikhanyayo

Ukukhutshwa kwe-GPT-5.2 kuza kunye nokutshintsha kweqhinga ngaphakathi kwe-OpenAI. Ngokwemithombo ekhankanyiweyo, umyalelo kaSam Altman ucacile: ukubeka imisebenzi emangalisayo ngasemva kwaye ugxininise kwinto exabiseke kakhulu ngabasebenzisi bamandla kunye nabathengi bequmrhu.

Oku kuguqulela ekunqumameni okanye ekucutheni amanyathelo afana nala iindlela ezintsha zokwenza imali kwi-ChatGPT, udibaniso lwentengiso, okanye ukusasazwa kwe ii-arhente ze-AI zexesha elide Ezi projekthi, zigxile kwiindawo ezinovakalelo ezifana nempilo okanye ukuthenga umntu siqu, zisesetafileni, kodwa zithathe indawo yokuhlala.

Eyona nto iphambili yangoku kukuqinisa ingundoqo yenkonzo: amava azinzileyo kunye nokucima okumbalwa, ukuphathwa okunamandla kwemithwalo enzima, kunye nemvakalelo epheleleyo yokuba imodeli "icinga ngcono" kwaye yenza iimpazamo ezimbalwa. Ngokwamagama abanye abahlalutyi, i-OpenAI ikhethe "Ingxolo encinci kunye nezihlunu ezininzi".

Le filosofi ihambelana neemfuno zenani elikhulayo leenkampani zaseYurophu ezidibanisa i-ChatGPT kwiinkqubo zabo zangaphakathi: ukusuka kwiijeneretha eziqulunqiweyo kunye nabancedisi benkxaso kwizixhobo zokuhlalutya amaxwebhu asemthethweni, obugcisa, okanye imali. Kwezi meko zokusetyenziswa, i ukuthembeka okuzinzileyo ngokuhamba kwexesha Ixabiseke ngaphezu kwayo nayiphi na intsebenzo enye yomboniso.

Kwinqanaba lemveliso, ezininzi zokuphuculwa kwe-GPT-5.2 kulindeleke ukuba ziqondwe njenge utshintsho cwakaAkusayi kubakho uhlengahlengiso olukhulu lwe-interface okanye "ubuso" obutsha be-chatbot, kodwa uhlengahlengiso lwangaphakathi olubonakalayo mihla le, kwindlela ephendula ngayo, kuthatha ixesha elingakanani, kwaye igcina kangakanani ukungaguquguquki.

Umjikelo wohlaziyo olukhawulezileyo

Ukuqaliswa kwe-GPT-5 ehlotyeni, kulandelwa yi-GPT-5.1 ngoNovemba, kwaye ngoku kufika ngokukhawuleza GPT-5.2 ngaphantsi kwenyangaBazoba ipateni ecacileyo: I-OpenAI ingene kumjikelo wokuphindaphinda okukhawulezayo, olungiselelwe yinkqubela phambili yayo kunye neentshukumo ezinobundlobongela zabakhuphisana nabo abanjengoGoogle kunye neAnthropic.

Kulo mongo, imihla yokukhupha iya isanda eguqukayoIzicwangciso zingatshintsha ngomzuzu wokugqibela ngenxa yemicimbi yeziseko ezingundoqo, uhlengahlengiso lokhuseleko, uvavanyo lomthwalo, okanye nezigqibo ezicwangcisiweyo zokuphendula izibhengezo zabakhuphisana nabo. Nangona kunjalo, ukunyanzeliswa kwe-9 kaDisemba njengento ekujoliswe kuyo ngaphakathi kubonisa ukuba uhlaziyo lusendleleni.

Kuluntu lwabasebenzisi kunye nabaphuhlisi baseSpain nakwiYurophu iphela, oku utshintsho olwenzekayo Inomphetho ophindwe kabini: kwelinye icala, iqinisekisa ukuba isixhobo sihlala siphucula; kwelinye, inyanzelisa ukuziqhelanisa nokusingqongileyo apho amandla emodeli anokutshintsha ngokuncomekayo kwisithuba seeveki.

Iinkampani, izitudiyo zophuhliso, kunye nee-arhente zikarhulumente eziye zaqala ukwakha iinkonzo kwi-OpenAI API kufuneka ziqwalasele indlela I-GPT-5.2 iguqula indlela yokuziphatha yemodelingakumbi kwimisebenzi enempembelelo ephezulu efana nokuveliswa kwekhowudi, uncedo lwezomthetho, okanye uhlalutyo lwedatha oluyimfihlo.

Ngelo xesha, uxinzelelo lokhuphiswano lukhuthaza bonke abadlali kwicandelo-OpenAI, Google, Anthropic, Meta-ukuqinisa umsebenzi wabo ukhuseleko, ukuthomalalisa umkhethe, kunye nolawulo lokubona izinto ezingekhoyo, imiba apho umlawuli waseYurophu ehlawula ingqalelo ekhulayo ngaphakathi kwesakhelo somthetho omtsha we-AI.

Impembelelo kubasebenzisi kunye namashishini eSpain naseYurophu

Nangona i-OpenAI ingekacacisi ukuba ukukhutshwa kwengingqi kuya kucwangciswa njani, kuqhelekile ukuba amanqaku amatsha e-ChatGPT kunye ne-API yakhe ifike phantse ngaxeshanye. EUnited States naseYurophu, kunye nomahluko omncinci ngokuxhomekeke kuhlobo lwenkonzo kunye nezibophelelo zolawulo lommandla ngamnye.

Kubasebenzisi ngabanye, Uphuculo olubonakalayo kwi-GPT-5.2 luya kuba yimvakalelo yobuchule obukhulu kunye nokunciphisa ukuthambekela kweempazamo. kwiingcaciso ezintsonkothileyo, izishwankathelo ezinde, okanye imisebenzi yokuyila enamanyathelo amaninzi. Ngamanye amagama, i-ChatGPT ebonakala ngathi "ibambekile" kancinci kwaye igcina ukuhamba kakuhle xa iincoko zinde.

Kwicandelo lezoshishino, ngakumbi kwiindawo ezifana nebhanki, amandla, unxibelelwano, kunye nokubonisana nobukho obuqinileyo eSpain kunye ne-EU, ukuzibophelela kwe-OpenAI isantya, uzinzo kunye nokwenza ngokwezifiso block nge imfuno yabancedisi bangaphakathi abanokusingatha imiqulu emikhulu yolwazi ngolongamelo olungephi lwezandla.

Ukuqiqa okuphuculweyo kukwangundoqo kumaqela e uphuhliso lwesoftware kunye nohlalutyo lwedathaEzi modeli zisetyenziselwa ukuphonononga ikhowudi, ukuphonononga iingcamango, okanye ukuvelisa amaxwebhu obugcisa. Izinga elisezantsi lokusilela kunye nengqiqo eyomelele ngakumbi inokuguqulela kwi... ukonga ixesha kunye nohlaziyo olumbalwa olulandelayo.

Kwelinye icala, isigqibo nqumama ukusasazwa kweearhente ze-AI ezizimeleyo, ngakumbi kwiindawo ezinobuthathaka njengezempilo, Oku kuhambelana nesilumkiso amaziko amaninzi aseYurophu ebesibiza.Phambi kokwenza umtsi kwiinkqubo ezenza izigqibo ezizimeleyo, iBrussels kunye nabalawuli belizwe bafuna iziqinisekiso ezingqongqo ekuhleni, ukubeka iliso ebantwini, kunye nokulandeleka.

Lilonke, i-GPT-5.2 ibumba ukuba ibe luhlaziyo oluyilelwe yomeleza ukuthembana kusetyenziso lobuchwephesha kunye nemihla ngemihla yemodeli, inkalo ebaluleke kakhulu kwiimarike apho umgaqo we-AI uqhubela phambili ngokukhawuleza kwaye iiprojekthi ezinkulu zihlalutywa ngekama lamazinyo.

Ukhuphiswano oluseta isantya sokwenza izinto ezintsha

Isigqibo se-OpenAI sokuvula "ikhowudi ebomvu" kunye nokuqhubela phambili kwe-GPT-5.2 sibonisa indlela ukhuphiswano noGoogle Ibeka isantya sokwenziwa kwezinto ezintsha kwi-AI namhlanje. Ukuba kwiminyaka embalwa edlulileyo yayikuvela kwe-ChatGPT eyanyanzela uGoogle ukuba asabele ngokukhawuleza, ngoku yiGemini 3 enyanzela i-OpenAI ukuba iphinde ihlele izinto eziphambili.

Oku kuguquguqukayo akuphelelanga ekuthengiseni: imodeli nganye entsha ephakamisa umgangatho wokusebenza inyanzela abanye ukuba bahlaziye iimephu zabo zendlela, baphinde banike amaqela, kwaye benze izigqibo ezinzima malunga nokuba zeziphi iiprojekthi ezimisiweyo nezithi zibe zezobuchule. Kule meko, umyalezo wangaphakathi kwi-OpenAI ngulo ukomelela kwemodeli yesiseko Inobunzima obungaphezulu kwayo nayiphi na imisebenzi eyongezelelweyo.

Kwi-ecosystem yetekhnoloji yaseYurophu, lo mzabalazo unomlinganiselo owongezelelweyo. Ukudityaniswa kwemifuziselo embalwa emikhulu yokulinganisa, elawulwa ziinkampani zase-US, kuhambisana neenzame zeManyano yaseYurophu yokukhuthaza iindlela zayo ezizezinye nokuqinisekisa ukuba ukuthunyelwa kwe-AI kuhambelana nemimiselo yayo. ubumfihlo, ukhuseleko kunye namalungelo asisiseko.

Ngeli xesha, zombini ezincinci zokuqala kunye neenkampani ezinkulu kwilizwekazi liphela Balandela ngokusondeleyo yonke intshukumo ukusuka OpenAI kunye Google, ukwazi ukuba amandla kunye nokunciphisa iimodeli zabo Ziya kuchaphazela iiprojekthi ezizenzekelayo, abancedisi benyani, uhlalutyo oluphambili, kunye nenkonzo yabathengi kwiinyanga ezizayo.

Kule meko iguquka ngokukhawuleza, i Ukukhutshwa kwangaphambili kwe-GPT-5.2 itolikwa njengenzame zokubuyisela ezinye zezinto ezifuziselayo kunye nezobuchwephesha ezisanda kulahleka kwi-Gemini 3.ukuqinisa uluvo lokuba idabi lobunkokeli be-AI ngoku liliwa iveki neveki, hayi kwimijikelo yonyaka.

Ngomhla we-9 kaDisemba njengosuku ekujoliswe kulo kunye nemephu yendlela egxile kwisantya, ukuqiqa kunye nokuthembeka, I-GPT-5.2 ijolise ekunikeni i-ChatGPT umva loo mvakalelo yemodeli eyomeleleyo kunye ne-agile ababeyilahlekile abasebenzisi abaninzi.Ngelixa uthumela umyalezo ocacileyo kukhuphiswano: I-OpenAI ayifuni ukuvumela uGoogle eyedwa ukuba asete isantya kwisizukulwana esilandelayo sobukrelekrele bokwenziwa.

Inqaku elidibeneyo:

Uyisebenzisa njani iMeta's MusicGen ekuhlaleni ngaphandle kokulayisha iifayile efini

UAlberto Navarro

Ndingumntu othanda itekhnoloji ojike umdla wakhe we "geek" waba ngumsebenzi. Ndichithe ngaphezulu kweminyaka eli-10 yobomi bam ndisebenzisa itekhnoloji yokusika kwaye ndikhenkceza ngazo zonke iintlobo zeenkqubo ngenxa yokufuna ukwazi okumsulwa. Ngoku ndiqeqeshelwe ubugcisa bekhompyutha nakwimidlalo yevidiyo. Oku kungenxa yokuba ngaphezu kweminyaka emi-5 ndibhalela iiwebhusayithi ezahlukeneyo kwitekhnoloji kunye nemidlalo yevidiyo, ndisenza amanqaku afuna ukukunika ulwazi oludingayo ngolwimi oluqondakalayo kuye wonke umntu.

Ukuba unayo nayiphi na imibuzo, ulwazi lwam lusuka kuyo yonke into enxulumene nenkqubo yokusebenza yeWindows kunye ne-Android yeefowuni eziphathwayo. Kwaye ukuzinikela kwam kukuwe, ndihlala ndikulungele ukuchitha imizuzu embalwa kwaye ndikuncede usombulule nayiphi na imibuzo onokuba nayo kweli lizwe le-intanethi.