Ungayisebenzisa kanjani i-OpenAI O1? Igayidi yomsebenzisi, intengo kanye nokubuyekezwa kokuhlangenwe nakho

Awukazi ukuthi uyisebenzisa kanjani i-OpenAI O1 ? Ngokuqinisekile "usemuva kwesikhathi"!

Kule nkathi yokuqhuma kobuhlakani bokwenziwa, i-OpenAI O1 ilethe ukusungulwa okusha kwezobuchwepheshe okungakaze kubonwe ngaphambili. Manje, ngizokuthatha ukuze uthole, ukuze ukwazi ukuguqula kusuka "ku-intelligence intelligence blind" kuya "kuchwepheshe we-AI" ngemizuzwana.

Iyini i-OpenAI O1?

I-OpenAI O1, igama lizwakala lisezingeni eliphezulu kakhulu, liwuhlelo olusha lwesizukulwane esisha lobuhlakani olwethulwe yi-OpenAI.

Kalula nje, i-O1 “iwubuchopho obukhaliphile” obuhlakaniphe futhi obusebenza kahle kakhulu kunanoma yisiphi isizukulwane sangaphambilini sobuchwepheshe be-AI. Kungakhathaliseki ukuthi ukugeleza komsebenzi okuzenzakalelayo noma ukucubungula kolimi lwemvelo oluyinkimbinkimbi, i-O1 ingakwazi ukuyiphatha kalula.

Akukona nje kuphela ukukongela isikhathi esiningi emsebenzini wakho wansuku zonke, kodwa futhi kungathuthukisa ukunemba kokwenza izinqumo futhi kukusize wenze ukukhetha okuhlakaniphile.

Ngakho-ke, nansi umbuzo: Ungayisebenzisa kanjani i-OpenAI O1? Ingakanani intengo? Kuthiwani ngokuhlangenwe nakho? Okulandelayo, ngizokuphendula ngamunye ngamunye.

Ungayisebenzisa kanjani i-OpenAI O1? Umhlahlandlela wokuqalisa osheshayo

Ungase ucabange ukuthi ukusebenzisa i-AI kuyinkimbinkimbi, kodwa empeleni, ukusebenza kwe-OpenAI O1 kulula kakhulu, futhi kungathiwa "ukusebenza njengesiwula". Landela lezi zinyathelo ezimbalwa futhi ungaqala kalula.

Indlela 1:ukubhalisaI-ChatGPT Futhi isebenzisa i-OpenAI O1

Isinyathelo sokuqala, vele, ukubhalisa nokungena ku-akhawunti ye-OpenAI.

• I-ChatGPT Plus kanye nabasebenzisi beThimba sebevele banokufinyelela ekuhloleni kuqala kwe-o1.

Isinyathelo sesibili ukubhalisela i-ChatGPT Plus.

• Okubhaliselwe kwe-ChatGPT Plus kubiza u-$20 ngenyanga.
• Lokhu kuvumela abasebenzisi ukuthi babeke phambili i-ChatGPT ngesikhathi samahora aphakeme futhi bathole izimpendulo ezisheshayo.

Uma ubhalisa i-OpenAI ezweni laseChina, ngokushesha "OpenAI's services are not available in your country."▼

Ngenxa yokuthi imisebenzi ethuthukisiwe idinga abasebenzisi ukuthi bathuthukele ku-ChatGPT Plus ngaphambi kokuthi isetshenziswe, kunzima ukwenza i-ChatGPT Plus isebenze emazweni angayisekeli i-OpenAI, futhi badinga ukubhekana nezinkinga ezinzima njengamakhadi esikweletu abonakalayo angaphandle...

Lapha sethula kuwe iwebhusayithi ethengeka kakhulu ehlinzeka ngama-akhawunti okuqasha okwabelwana ngawo e-ChatGPT Plus.

Sicela uchofoze isixhumanisi esingezansi ukuze ubhalise i-akhawunti yePlanet ▼

Chofoza isixhumanisi esingezansi ukuze uthole umhlahlandlela onemininingwane wokubhalisa i-akhawunti yakho ku-Planet▼

🔗Xhumanisa ne-ChatGPT Plus ukuthenga nge-carpool yabantu abaningi kwi-akhawunti yeplanethi, thola isaphulelo esingu-50% emalini yenyanga!

Amathiphu:

• Amakheli e-IP e-Russia, e-China, e-Hong Kong, nase-Macau awakwazi ukubhalisela i-akhawunti ye-OpenAI. Kunconywa ukuthi ubhalise elinye ikheli le-IP.

Indlela yesi-2: Sebenzisa i-OpenAI O1 ngokubiza isixhumi esibonakalayo se-API

I-OpenAI O1 inokuxhumana okuhlukahlukene kwe-API futhi inezixazululo ezihlukene zezidingo ezahlukene.

• Isibonelo, uma ungunjiniyela, ungakhetha isixhumi esibonakalayo se-API esiqondene nonjiniyela;
• Uma ungumsebenzisi webhizinisi, i-Enterprise Edition API ingakusiza ukuthi uphathe kangcono ukuhamba komsebenzi kwangaphakathi kwenkampani yakho.

 Imingcele yokumisa

Ungesabi ubuchwepheshe! Ngisho noma ungazi lutho ngobuchwepheshe, i-interface ye-O1 yakhelwe ukuthi isebenziseke kalula. Udinga kuphela ukulandela ukwaziswa kusixhumi esibonakalayo bese ufaka imingcele oyidingayo, njengohlobo lwedatha yolimi okufanele lucutshungulwe, isitayela sokwenziwa kwemibhalo, njll., futhi uhlelo luzokhiqiza ngokuzenzakalelayo isisombululo esifanele ngokusekelwe kulolu lwazi. .

Qalisa umsebenzi futhi uthole imiphumela

Lapho usufake wonke amapharamitha adingekayo, chofoza "Run". I-O1 izocubungula ngokuzenzakalelayo imisebenzi oyithumelayo, noma ngabe ikhiqiza umbhalo, ihlaziya idatha, noma eminye imisebenzi eyinkimbinkimbi, futhi ingaqedwa emizuzwaneni embalwa. Imiphumela izovezwa ngokuzenzakalelayo ku-akhawunti yakho yasemuva, futhi ungaqhubeka ulungisa noma ulungiselele njengoba kudingeka.

 Thuthukisa futhi wenze ngokugcwele

Ucabanga ukuthi ukukhishwa kwe-O1 akulindelekile? Ungakhathazeki, i-O1 isekela ukulungiselelwa okuphindaphindiwe. Kancane kancane ungenza i-AI "ikuqonde" ibe ngcono ngokulungisa amapharamitha noma ngokunikeza amasampuli okokufaka amaningi kuyo. Uma uyisebenzisa isikhathi eside, umphumela uzoba ngcono.

Ibiza malini i-OpenAI O1? Ifanele abantu ngabanye noma amabhizinisi?

Uchungechunge lwemodeli ye-OpenAI ye-o1 luhlanganisa izinhlobo ezimbili: i-o1-preview kanye ne-o1-mini. Zithuthukisiwe ekhonweni lokucabanga, ikakhulukazi emkhakheni wesayensi , ubuchwepheshe, ubunjiniyela, kanye nezibalo (STEM). Imodeli ye-o1-preview ihamba phambili ekuhlinzekeni ububanzi obubanzi "bolwazi lomhlaba," kuyilapho i-o1-mini ingcono kokuqukethwe kokucabanga njengokubhala ikhodi, kodwa ingase ingabi namandla kangako ngokolimi nolwazi olujwayelekile.

在价格方面，o1-preview模型的API调用价格为每输入100万个token 15美元，每输出100万个token 60美元。相比之下，o1-mini模型的价格较为经济，其费用为每输入100万个token 3美元，每输出100万个token 12美元，这比o1-preview模型便宜了80%。

o1模型的使用限制相对较低，目前允许o1-preview每周使用30次，o1-mini每周使用50次。这些限制可能会随着用户需求和反馈逐步提升。

I-OpenAI ihlela ukwethula inguqulo yomongo enkudlwana yemodeli ye-o1 esikhathini esizayo futhi kancane kancane ivule i-o1-mini kubo bonke abasebenzisi bamahhala be-ChatGPT, nakuba isikhathi esithile singakanqunywa. Ngesikhathi esifanayo, i-OpenAI iphinde isebenzele ukwandisa imikhawulo yokusetshenziswa ngokuhamba kwesikhathi, futhi ihlela ukwengeza imisebenzi efana nokuphequlula iwebhu, ifayela nokulayishwa kwezithombe, njll., ukwenza imodeli ibe usizo kakhulu. Mayelana nokuthambekela kwesikhathi esizayo kwamanani entengo, i-OpenAI ithe ngokomlando, amanani ehle izikhathi ezingu-1 njalo eminyakeni engu-2-10, futhi lo mkhuba ungase uqhubeke.

Uma sikhuluma ngentengo, lo ngumbuzo wonke umuntu akhathazeke ngawo. Phela, ukuthi ungakwazi "ukuchitha imali encane ukwenza izinto ezinkulu" kuhlobene ngokuqondile nenani lalo mkhiqizo.

1. Inguqulo yesilingo samahhala

Ukuze uvumele wonke umuntu abe nesipiliyoni esingcono se-O1, i-OpenAI ingase inikeze inguqulo yesilingo samahhala.

Kubasebenzisi abajwayelekile, nakuba i-o1-mini ingavuliwe okwamanje, kulindeleke ukuthi ivuleleke kuwo wonke umuntu esikhathini esizayo esiseduze.

Nakuba imisebenzi inomkhawulo, kwanele ukuthi uzwe ukusebenza kwayo okunamandla kanye nesixhumi esibonakalayo esilula sokusebenza. Uma uyisebenzisa ngezikhathi ezithile kuphela, noma ufuna nje ukuthola ukuqonda kokuqala kwezici ze-O1, le nguqulo yanele.

2. Inguqulo eyisisekelo

I-ChatGPT Plus kanye nabasebenzisi beThimba sebevele banokufinyelela ekuhloleni kuqala kwe-o1.

Okubhaliselwe kwe-ChatGPT Plus kubiza u-$20 ngenyanga. Lokhu kuvumela abasebenzisi ukuthi babeke phambili i-ChatGPT ngesikhathi samahora aphakeme futhi bathole izimpendulo ezisheshayo.

Le nguqulo ifanele amabhizinisi amancane nabasebenzisi ngabanye, futhi imisebenzi yayo ihlanganisa izidingo zansuku zonke, njengokukhiqiza umbhalo, ukuhlaziywa kwedatha, njll. Inzuzo yayo ilele ekuguquguqukeni kwayo okuphezulu futhi ifanele abasebenzisi abadinga ukusetshenziswa njalo kodwa abanesabelomali esilinganiselwe.

3. Uhlelo lwebhizinisi

Inguqulo yebhizinisi yakhelwe izinkampani ezinkulu neziphakathi Intengo iphakeme kakhulu, kodwa imisebenzi nayo inamandla ngokwedlulele. Akukhona nje kuphela ukuthi ingakwazi ukuphatha inani elikhulu ledatha, futhi ingathuthukisa ngokugcwele futhi ithuthukise ukuhamba komsebenzi kwenkampani. Uma uyinkampani edinga ukuthuthukisa ukukhiqiza nokuthuthukisa amakhono okucubungula idatha, le nguqulo kufanele nakanjani utshale kuyo.

Kukuhambela kanjani ukusebenzisa i-OpenAI O1? Isibuyekezo silapha!

Njengomsebenzisi wesipiliyoni esijulile, kufanele ngisho ukuthi ukusebenza kwe-OpenAI O1 "kuyamangaza".

Ihlakaniphile futhi iyashesha kunanoma yiluphi uhlelo lwangaphambilini lwe-AI mayelana nokukhiqiza umbhalo ohlakaniphile, ukuhumusha, nokuhlaziywa kwedatha.

1. Isivinini nokusebenza kahle

Mayelana nesivinini, isikhathi sokuphendula se-O1 sifushane kakhulu. Kungakhathaliseki ukuthi uhambisa muphi umsebenzi, ungaqedwa ngemizuzwana. Ikakhulukazi emisebenzini edinga ukucubungula amanani amakhulu edatha, ukusebenza kwe-O1 kuvelele kakhulu. Uma kuqhathaniswa nezinye izinhlelo ze-AI, isebenza kahle okungenani ngo-30%.

2. Ukunemba kanye nobuhlakani

I-O1 ayisheshi nje kuphela, kodwa futhi "inembile". Iziqonda ngokunembile izidingo zabasebenzisi futhi ikhiqiza ngokuzenzakalelayo okuqukethwe kokuphumayo kwekhwalithi ephezulu. Mayelana nokucutshungulwa kolimi lwemvelo, i-O1 ingabona ngokuzenzakalela izimo ezihlukene futhi ikhiqize umbhalo ofanele noma imiphumela yokuhumusha ngokusekelwe kumongo.

3. Isixhumi esibonakalayo somsebenzisi

Idizayini yesixhumi esibonakalayo ye-O1 ilula futhi icacile, okwenza kube lula ngisho nakubantu abangenazo isizinda sobuchwepheshe ukuqalisa. Kuneziqondiso ezicacile namadokhumenti osizo kuso sonke isinyathelo, ngakho-ke azikho izithiyo ohlelweni lokusebenzisa.

4. Ukuqina

Kubasebenzisi bebhizinisi, okugqama kakhulu kwe-O1 ukuqina kwayo. Ungathuthukisa imisebenzi futhi wandise i-API nganoma yisiphi isikhathi ngokuya ngezidingo zebhizinisi. Ngisho noma izidingo zakho zishintsha, i-O1 ingakwazi ukuzivumelanisa nezimo ngokushesha, ikugcine uphambili emqhudelwaneni.

o1 uchungechunge lwenguqulo isingeniso

Uchungechunge lwe-o1 lunezinguqulo ezimbili: i-o1-mini kanye ne-o1-preview.

o1-ukubuka kuqala

• Lokhu ukubuka kuqala kwemodeli ye-o1 esemthethweni ephezulu ezayo.
• i-o1 izuze impumelelo ebalulekile ekucabangeni kobuhlakani bokwenziwa.

o1 mini

• Lena imodeli esheshayo neshibhile eyenza kahle kakhulu emisebenzini yokuhlela.
• Njengenguqulo encane, i-o1-mini inenani kuphela ekubukeni kuqala kuka-o1 20%, kuyilapho silondoloza ukucabanga okuphumelelayo futhi singonga kakhulu futhi sithengeka.

Amandla okucabanga e-OpenAI O1 empeleni asebenza kanjani

I-OpenAI ikhomba ngqo ukuthi la mamodeli amasha, aqeqeshwe ngokufunda okuqinisiwe , ayakwazi ukusingatha imisebenzi yokucabanga eyinkimbinkimbi.

Ngakho-ke, ngokwesimo semodeli yolimi enkulu, kusho ukuthini ngempela "ikhono lokucabanga" ?

Njengoba nje abantu kudingeka bacabange ngokujulile lapho bebhekene nezinkinga ezinzima, u-O1 usebenzisa "uchungechunge lwemicabango" lapho exazulula izinkinga.

• Kancane kancane ihlephula izinyathelo eziyinkimbinkimbi zibe izingcezu ezilawulekayo ngokukhomba nokulungisa amaphutha;
• Uma isu lamanje lihluleka, lizozama imibono emisha.

Inqubo yokucabanga ihlanganisa lezi zinyathelo ezilandelayo:

1. Khiqiza amathokheni okucabanga；
2. Omaka bempendulo ebonakalayo okukhiphayo；
3. Susa omaka bezinkomba kumongo,

Gcina umongo ugxile emyalezweni owumongo ngokususa omaka bezinkomba.

Kubalulekile ukuqaphela ukuthi nakuba omaka bezinkomba bengabonakali ku-API, basahlala endaweni yewindi lomongo wemodeli futhi bafakiwe ezindlekweni zomaka okukhiphayo.

Imibono kaJim Fan

Umcwaningi omkhulu we-NVIDIA uJim Fan ukhombe ukuthi nakuba le nqubo yokunquma ingase ihambe kancane, indlela yokunquma isikhathi eside ekugcineni isithandwa kakhulu ezindaweni zokukhiqiza.

UJim uphinde wanikeza imininingwane eminingana ethokozisayo:

1. Ukucabanga okuhlakaniphile akuncikile kumamodeli amakhulu : amapharamitha amamodeli amaningi amakhulu asetshenziswa kakhulu ukugcina ulwazi oluyiqiniso ukuze kusingathwe izinkinga ezivamile. Kodwa-ke, singahlukanisa amakhono okucabanga ezisekelweni zolwazi. Cabanga "ngomqondo wokucabanga" oyinkimbinkimbi okwazi ukubiza amathuluzi ahlukahlukene ngendlela eguquguqukayo (njengokusesha kuwebhu noma amathuluzi okuhlola ikhodi) ukuze uthole ulwazi oludingekayo. Isu elinjalo linganciphisa kakhulu izinsiza zokubala ezidingekayo ekuqeqeshweni kwe-AI.

2. Imodeli entsha ishintsha inani elikhulu lomsebenzi wokubala kusukela esigabeni sokuqeqesha kuya esigabeni sokusetshenziswa okusebenzayo. Ungacabanga ngemodeli enkulu yolimi "njengomhlaba olingiswayo" osekelwe embhalweni. Lapho ixazulula inkinga, imodeli izama izindlela nezimo ezahlukahlukene ngaphakathi kwaleli "zwe elilingiswayo" ukuthola ikhambi elifanele. Le nqubo ifana kancane nesofthiwe yokudlala i-chess ( njenge -AlphaGo) elilingisa ukunyakaza okuningi engqondweni yayo ukuze inqume ukunyakaza okulandelayo okuhle kakhulu.

Ukuqhathaniswa kwe-o1 ne-GPT-4o

Ukuze kuhlolwe ukusebenza kwemodeli ye-o1 uma kuqhathaniswa ne -GPT-4o , i-OpenAI yenze ukuhlolwa okubanzi kanye nezilinganiso zokufunda komshini.

Imiphumela ikhombisa ngokusobala ukuthi i-o1 isebenza kahle kakhulu kune-GPT-4o emisebenzini yokucabanga eyinkimbinkimbi njengezibalo , izinhlelo, kanye nezinkinga zesayensi .

Ikakhulukazi, i-o1 yenze kahle kakhulu ekuhlolweni kwe-GPQA-diamond benchmark. Lokhu kuhlolwa kwenzelwe ukuhlola ubuchwepheshe bemodeli emikhakheni yekhemistri, ifiziksi, kanye nebhayoloji .

Ukuqhathanisa ukusebenza kwemodeli nabantu, i-OpenAI imeme abanikazi be-PhD ukuthi baphendule imibuzo efanayo ye-GPQA-diamond.

Imiphumela eshaqisayo

I-o1 idlule laba chwepheshe, yaba imodeli yokuqala ye-AI ukwedlula ama-PhD kuleli bhentshimakhi. Yize lokhu kungasho ukuthi i-O1 ingcono kune-PhD kuzo zonke izindawo, kukhombisa ukuthi inekhono kakhulu ekuxazululeni izinkinga ezithile.

Ungabuka umbiko wobuchwepheshe kumodeli we-O1 lapha .

Isibonelo sokuqhathanisa: i-o1 ne-GPT-4o

Ake siqhathanise ukusebenza kwe-o1 nemodeli yangaphambilini ye-GPT-4o ngombuzo wakudala: Bangaki "ama-r" akhona egameni elithi "isitrobheli"?

Prompt: How many ‘r’ letter are in the word strawberry?

• o1 esetshenzisiweImizuzwana engu-33, esetshenzisiwe296 amamaki, wanikeza impendulo enembile;
• I-GPT-4o ayisebenzisi kuphelaImizuzwana engu-1, ukusetshenziswa39 amamaki, kodwa wehlulekile ukuphendula ngokunembile.

Zama omunye umbuzo:

Zombili izinhlobo zacelwa ukuthi zibhale amazwe amahlanu amagama awo anohlamvu "A" endaweni yesithathu.

Prompt: Give me 5 countries with letter A in the third position in the name

• i-o1 iphinde yanikeza impendulo enembile, nakuba kuthathe isikhathi eside "ukucabanga" kune-GPT-4o.

Imicabango yami nge-OpenAI O1: Ithuluzi lenguquko noma i-gimmick?

Kafushane, i-OpenAI O1 "iyithuluzi lokuguqula". Akwenzi kube lula kakhulu ukucutshungulwa kwemisebenzi eyinkimbinkimbi, kodwa futhi kwenza ukusetshenziswa kwe-AI kudume futhi kusebenze. Kubasebenzisi abangabodwana, i-O1 ingumsizi omuhle wokuthuthukisa ukusebenza kahle nokunciphisa umthwalo womsebenzi kubasebenzisi bebhizinisi, ilethe inguquko yokukhiqiza engakaze ibonwe.

Kodwa-ke, kufanele kuqashelwe ukuthi i-O1 ayiyena umninimandla onke. Nakuba umsebenzi wayo unamandla kakhulu, udinga izindleko ezithile zokufunda kanye nenqubo yokuzivumelanisa nezimo. Uma kungokokuqala ngqa usebenzisa ithuluzi le-AI, kungase kuthathe isikhathi ukuqonda indlela yokusebenza nokusebenza kwayo.

Isifinyezo: Kungani udinga i-OpenAI O1?

I-OpenAI O1 ingaphezu nje kwethuluzi le-AI, inguzakwethu ongcono kakhulu wokuthuthukisa ukusebenza kahle nokuthuthukisa ibhizinisi. Noma ngabe ungumsebenzisi oyedwa noma umenzi wezinqumo wenkampani, i-O1 ingakusiza ukuthi uthole umphumela ophindwe kabili ngohhafu womzamo emsebenzini wakho wansuku zonke.

Ngakho-ke, uma usangabaza, ungase futhi ubambe ithuba lokuzama i-OpenAI O1 futhi uzwe amandla okuguqula ewalethayo. Khumbula, kule nkathi yokuqhuma kolwazi, yilabo kuphela "abasebenzisa kahle i-AI" abangakwazi ukunqotshwa.

Hamba uyohlola i-OpenAI O1 futhi uqale inkathi entsha yomsebenzi ophumelelayo kuwe!