Imodeli ye-OpenAI ezayo ye-Astra idala ukunakekela ngamakhono ayo athuthukile okubhala amakhodi namakhono okuphepha kwe-inthanethi, kuyilapho imibiko isikisela ukuthi ingase futhi ikhiqize izixhumi eziyinkimbinkimbi nokokuzithokozisa okusebenzisanayo kwe-3D ngemiyalelo elula. Imodeli kulindeleke ukuthi iqhubekisele phambili i-AI ekwakhiweni kwesofthiwe ezimele, nakuba imininingwane eminingana ephathelene namakhono ayo kanye nokugcinwa kwebhrendi ingakaqinisekiswa.
I-OpenAI ilungiselela ukukhulula i-Astra, imodeli yobuhlakani bokufakelwa ezayo engamelela enye yentuthuko enkulu yenkampani ekufakeni amakhodi, ukuphepha kwe-inthanethi nokuxazulula izinkinga ngokuzimela.
I-Astra iye yaxhunyaniswa kakhulu emibikweni eya esizukulwaneni esilandelayo se-ChatGPT ne-GPT-6. Eminye imibiko ichaze imodeli ngokuthi “ChatGPT 6 Astra” futhi ithi ihlolwe ngaphansi kwegama elithi “Ultima Alpha.“ I-OpenAI, nokho, ibhekisele ngokusemthethweni ku-Astra njengemodeli nje ezayo futhi ayikaqinisekisi ukuthi izokwethulwa ngaphansi kwegama le-GPT-6.
Injabulo ezungeze i-Astra idlulela ngalé kwesizukulwane sombhalo esivamile. Imibiko isikisela ukuthi imodeli ingakhiqiza okokuzithokozisa okuyinkimbinkimbi kwedijithali, okuhlanganisa imidlalo ye-3D esebenzisanayo, izindawo zesitayela se-voxel kanye nezixhumi zomsebenzisi eziyinkimbinkimbi ngemiyalelo elula.
Uma lawa makhono efinyelela umkhiqizo wokugcina, abathuthukisi bangase basuke ekuchazeni umbono baye ekukhiqizeni okokuzithokozisa okusebenzisanayo okusebenzayo ngokushesha okukhulu kunokuhamba komsebenzi wokuthuthukisa wendabuko.
Ukuthuthukiswa kuzomelela esinye isinyathelo ekuguqukeni kwamathuluzi wokufaka amakhodi we-AI. Amamodeli we-AI wamanje kakade angakhiqiza izinhlelo zokusebenza, ahlaziye isofthiwe futhi asebenze kuwo wonke amakhodi amakhulu. Amasistimu e-agent anekhono kakhudlwana angakuqhubekisela phambili lokhu ngokuhlela imisebenzi, ngokusebenzisa amathuluzi okuthuthukisa nokuqedela amaphrojekthi ayinkimbinkimbi ngokungenelela kwabantu okungaqondile.
I-OpenAI iqinisekise ngokusemthethweni ukuthi i-Astra imelela intuthuko enkulu ekubhaleni amakhodi e-ejenti nasekuvikelekeni kwe-inthanethi. Le nkampani ithi ukuhlolwa kwayo kubonisa ukuthi imodeli ikwazi kakhulu ukwedlula i-GPT-5.6 Sol ekuboneni ubuthakathaka nasekuthuthukiseni ukuxhaphaza kuyilapho futhi idinga amathokheni ambalwa okukhiphayo ukuze ifeze leyo misebenzi.
I-Astra nayo ifinyelele ingqophamlando ebalulekile yezokuphepha kwe-inthanethi. I-OpenAI ikuhlanganise njengokuhlangabezana nomkhawulo obalulekile wekhono lokuphepha kwe-inthanethi ngaphansi koHlelo lwayo Lokulungiselela, okwenza kube imodeli yokuqala ye-OpenAI ukuthola leso sikhundla.
Ngaphansi kwaleso sigaba, imodeli ehlonyiswe ngokwanele ingakhomba ubuthakathaka bokuphepha obungaziwa ngaphambili futhi isungule izindlela ezisebenzayo zokuxhaphaza amasistimu ekhompyutha avikelwe ngaphandle kokudinga umuntu ukuba aqondise zonke izinyathelo zomuntu ngamunye.
Ngesikhathi sokuhlola, i-Astra ithole amaphuzu aphelele ekuhlolweni kwe-OpenAI kwe-ExploitBench ngokuthuthukisa ukuxhaphaza ebuthakathakeni obaziwayo. Inkampani iphinde yayihlola ngokumelene nesilinganiso sangaphakathi esihlukile esinobuthakathaka obuphakeme obusanda kudalulwa.
Okubaluleke nakakhulu, i-Astra yathola futhi yasebenzisa ubuthakathaka obubili obungaziwa njengengxenye yokuxhaphaza ngesikhathi sokuhlola. I-OpenAI ithi isebenza ukudalula lobobuthakathaka kubalondolozi besofthiwe abathintekile.
Imiphumela ibonisa indlela amakhono wokufaka amakhodi we-AI anda ngokushesha ngayo ngaphezu kokukhiqiza amazwibela esofthiwe. Amamodeli athuthukile akwazi ngokwengeziwe ukuzulazula ezindaweni zekhompyutha, ukubona ubuthakathaka, ukuhlanganisa amathuluzi nokuqedela ukulandelana kwesikhathi eside komsebenzi wezobuchwepheshe.
Kubathuthukisi, amakhono afanayo ekugcineni angashintsha indlela isofthiwe esungulwa ngayo. Esikhundleni sokwakha zonke izingxenye zohlelo lokusebenza, abasunguli bangachaza ngokwengeziwe lokho abafuna ukukwakha futhi bavumele ama-ejenti e-AI ukuthi asingathe izingxenye ezinkulu zokuqalisa, ukuhlola nokulungisa iphutha.
Abaklami bangabhekana noshintsho olufanayo. Uma imibiko ephathelene nesixhumi esibonakalayo se-Astra namakhono esizukulwane se-3D ibonakala inembile, ukukhiqiza ama-prototypes kungashesha kakhulu. Umthuthukisi noma umklami angase achaze isikhombimsebenzisi, indawo esisebenzisanayo noma umqondo womdlalo futhi athole indawo yokuqalisa esebenzayo phakathi nemizuzu.
Amakhono anjalo futhi aletha izingozi ezinkulu.
I-OpenAI ibambezele okwesikhashana izingxenye zentuthuko ye-Astra ngenkathi iqinisa izivikelo ezizungeze imodeli. Inkampani yethule izilawuli zokuvikeleka eziqine kakhudlwana, inethiwekhi namathuluzi akhawulelwe, amasistimu okuqapha athuthukisiwe nezivikelo eziqinile eziklanyelwe ukuvimbela imodeli ekwenzeni izenzo ezingagunyaziwe.
Lezi zinyathelo zokuqapha zibaluleke ngokukhethekile ngoba amakhono afanayo avumela i-AI ukuba ithole ubuthakathaka ngezinjongo zokuzivikela angase asetshenziswe kabi ekuhlaselweni kwe-cyber.
Ngakho-ke i-OpenAI ihlela ukukhawulela ukufinyelela ekusebenzeni okuthuthuke kakhulu kokuphepha kwe-inthanethi kwe-Astra. Abahloli abakhethiwe bazothola ukungena kusenesikhathi, futhi kulindeleke ukungena okubanzi kokuvikeleka kwe-inthanethi ngohlelo lwenkampani lwe-Daybreak Blue. Imodeli ejwayelekile ye-Astra kulindeleke ukuthi itholakale maduzane, nakuba i-OpenAI ingakamemezeli usuku oluqondile lokukhishwa.
Lokhu kuhlukile emibikweni esikisela ukukhishwa kuka-September 2026 kwemodeli ebizwa ngokuthi i-GPT-6 Astra. Nakuba ukuqaliswa okunjalo kusengenzeka, i-OpenAI ayikakuqinisekisi esidlangalaleni ukufaka uphawu lwe-GPT-6 noma usuku oluqondile lokukhishwa.
I-Astra iyafika njengoba ukuncintisana phakathi kwabathuthukisi be-AI abahamba phambili kuqhubeka kuqina. Ukufaka amakhodi sekungenye yezindawo ezibaluleke kakhulu ekuthuthukisweni kwe-AI ngoba isofthiwe inikeza amamodeli indlela yokuguqula ukucabanga kube yizenzo ezisebenzayo.
Ukuthuthuka kusuka ekugcwalisweni kwekhodi kuya kuma-ejenti wokufaka amakhodi okuzimela sekushintsha kakade ukuhamba komsebenzi wokuthuthukisa. Esikhundleni sokusebenzisa i-AI ukuze nje basikisele umugqa olandelayo wekhodi, abathuthukisi baya ngokuya besebenzisa ama-ejenti ukuhlola izinqolobane, ukuguqula amafayela amaningi, ukwenza imiyalo, ukuxilonga amaphutha nokusebenzisa izici eziphelele.
Amakhono e-Astra asikisela ukuthi lolu shintsho lungashesha ngokwengeziwe.
Ngakho-ke ukubaluleka kwesikhathi eside kungase kudlulele ngalé kokuthi imodeli igcina inegama le-GPT-6. I-Astra inikeza enye inkomba yokuthi amasistimu we-AI asemngceleni athuthukela ekubeni ama-ejenti ezobuchwepheshe azimele akwazi ukwenza imisebenzi eyinkimbinkimbi ngokwengeziwe ngaphakathi kwezindawo zangempela zekhompyutha.
Kubathuthukisi, abaklami namabhizinisi, lokho kunganciphisa kakhulu isikhathi esidingekayo sokuguqula imibono ibe imikhiqizo yedijithali esebenzayo. Kochwepheshe bezokuphepha kwe-inthanethi, ubuchwepheshe obufanayo bunganikeza amathuluzi amasha anamandla okuthola ubuthakathaka ngaphambi kokuba abahlaseli babuthole.
Inselele ye-OpenAI kanye nomkhakha obanzi we-AI kuzoba ukuqinisekisa ukuthi lawo makhono ahlala ewusizo ngaphandle kokufinyeleleka okuyingozi ngezinjongo ezinonya.
I-Astra ihlezi ngokuqondile kuleyo mpambano. Ithembisa ukuthuthukiswa okusheshayo, ukufakwa kwamakhodi okuzimele kakhudlwana kanye nezinhlobo ezintsha zokuqukethwe okusebenzisanayo okwenziwe nge-AI, kuyilapho ngesikhathi esifanayo kubonisa ukuthi kungani amasistimu e-AI akwazi ngokwengeziwe adinga izivikelo eziqinile.
I-OpenAI ithi imininingwane eyengeziwe mayelana nokuphepha kwe-Astra, ukuvumelanisa nokuhlolwa kokuvikeleka izoshicilelwa ekhadini lesistimu yemodeli lapho yethulwa. Kuze kube yileso sikhathi, izimangalo zebhrendi ye-GPT-6, isizukulwane se-3D nezinye izici ezingakhishwanga kufanele zithathwe njengemibiko kunokucaciswa okuqinisekisiwe.
Okuqinisekisiwe kakade kubaluleke ngokwanele. I-Astra ikwazi kakhudlwana ukuphepha kuyi-cyber kunezinhlobo zomphakathi ze-OpenAI zamanje, ingazitholela ngokuzenzakalelayo izingozi ezingaziwa ngaphambilini futhi ibe imodeli yokuqala ye-OpenAI ukudlula umkhawulo obalulekile wokuphepha kuyi-cyber. Lokho kwenza ukukhululwa kwayo ekugcineni kube okunye kokwethulwa kwe-AI okuwumphumela omkhulu kakhulu ukubukwa.