Wonke amathuluzi e-AI
I-ejenti ye-AI
Lapha kuhlanganiswe imikhiqizo ye-AI ethandwayo emkhakheni wama-ejenti e-AI, ehlanganisa ukudala, ukucubungula nokuthuthukisa ukusebenza kahle.
Qwen IntelligenceI-ejenti ye-AII-Qwen Intelligence iyisixazululo esiphelele sezingcingo ze-AI esethulwe ithimba le-Alibaba Tongyi Qianwen, esinikeza abakhiqizi bezingcingo amamodeli, inkundla nezixazululo ezisekelwe ezimweni zokusetshenziswa. I-Qwen Intelligence iqukethe ama-Agent amathathu amakhulu: i-Mobile Pla...00
Xiaomi MiMo-V2.6I-ejenti ye-AII-Xiaomi MiMo-V2.6 iwuchungechunge lwamamodeli e-multimodal olukhishwe futhi lwaba yi-open source yi-Xiaomi, oluqukethe amamodeli amabili e-multimodal omdabu, i-Pro ne-Flash, olugxile ku-Agentic reinforcement learning yezinga elikhulu egxile emisebenzini eyinkimbinkimbi engaqinisekiswa.00
Step 5 PreviewI-ejenti ye-AII-Step 5 Preview iyimodeli eyisisekelo ephambili yesizukulwane esisha eyethulwe yi-StepFun, eyakhelwe imisebenzi ye-Agentic yomhlaba wangempela. Imodeli isebenzisa ukwakheka kwe-sparse MoE, inamapharamitha angu-600B isiyonke, kanti kuvulwa angu-27B kuphela, futhi isekela okungaphezu kwezigidi zama-Token ...00
HypitI-ejenti ye-AII-Hypit iyithuluzi le-AI lokukhiqiza amavidiyo eliwumthombo ovulekile. Umqondo oyinhloko uwukunikeza i-Agent ividiyo edume kakhulu, bese i-Agent iyihlaziya ngokuzenzakalelayo iphinde idale kabusha yonke inqubo yayo yokusebenza. Ithuluzi libhala ividiyo njengefayela elihlelekile le-.svml; shintsha umbhalo owodwa, umsindo, ividiyo nemibhalo engezansi kuzivumelanisa ngokuzenzakalelayo.00
NeoHorse-1I-ejenti ye-AII-NeoHorse-1 iyimodeli ye-Agent yomthombo ovulekile ekhishwe ngokubambisana yi-Jiyuan Rhythm, i-Wuwen CoreSky, i-Tsinghua, i-Peking University, i-Chinese University of Hong Kong, i-Alibaba nezinye izinhlangano, futhi iza ngezinguqulo ezimbili: i-4B ne-9B. Imodeli isebenzisa ukuqeqeshwa kwangemva kwe-Agent-Native, isebenzisa imizila yokwenziwa yangempela ye-Routing ...00
Ukugcwaliswa kabusha kobulungu be-ChatGPT Plus ne-Pro 5X/20XI-ejenti ye-AIVipCheap - isitolo sokuqala sezaphulelo sezinzuzo zobulungu be-AI, esihlanganisa ukugcwaliswa kabusha okuqondile okusemthethweni kwamathuluzi e-AI adumile afana ne-ChatGPT Plus ne-Claude Pro00
MuseI-ejenti ye-AIUMuse uyi-Personal AI Agent eyethulwe nguMeta, eqonda izinhloso ngaphakathi kwemvume oyinikezile, ihlele futhi iqhubekisele phambili imisebenzi yansuku zonke ezinhlelweni ezahlukene, bese ibuyisela izinqumo ezibalulekile kumsebenzisi ukuze aziqinisekise.00
WikiSkillI-ejenti ye-AII-WikiSkill iwuhlaka lokuguquka kwamakhono e-Agent olwenziwe yi-Google Research. Lolu hlaka lusebenzisa ukwakheka kwezingqimba ezintathu: i-Raw Layer egcina imikhondo eluhlaza, i-Wiki Layer eqhubeka iqongelela ulwazi oluhlelekile, kanye ne-Skills Layer egcina amakhono asebenzisekayo...00
I-问卷派I-ejenti ye-AII-问卷派 (WeJot) iyi-ejenti yocwaningo eyakhiwe ngobuchwepheshe be-AI Agent. Umsebenzisi udinga kuphela ukuchaza inhloso yocwaningo ngolimi lwemvelo, bese i-AI iqedela yonke inqubo kusukela ekwenzeni uhlu lwemibuzo, ekuklanyweni kwemithetho eyinkimbinkimbi, ekufaniseni amasampula ahlakaniphile, ezingxoxweni ze-AI, ...00
Ama-Agent e-Claude CommerceI-ejenti ye-AIAma-Agent e-Claude Commerce ayithuluzi le-e-commerce elivulwe yi-Anthropic njenge-open source, elihlanganisa i-Agent yokuthenga ne-Agent yabathengisi. Amathuluzi ahlanganisa izimboni ezine: ukudayisa, ezokuvakasha, ezokuxhumana kanye namathikithi, anamasango okuphepha anezigaba eziningi, okuhlanganisa konke okubhalwayo okubanjwa okwesikhashana...00
PatentClawI-ejenti ye-AII-PatentClaw iyithuluzi elisiza elinomthombo ovulekile elenzelwe abacwaningi, abasunguli nabameli bamalungelo obunikazi, elingahlela ngokuzenzakalela amaphepha ocwaningo e-LaTeX abe imibhalo yezincazelo zobuchwepheshe yamalungelo obunikazi okusungula ase-China. Leli thuluzi lisebenzisa ukubambisana kwama-Agent amaningi ukuqedela ukuhlaziywa kwephepha...00
EaselI-ejenti ye-AII-Easel iyindawo yokusebenzela yokuqukethwe kwezokuxhumana esebenzisa i-AI, ekhishwe ngomthombo ovulekile ngokubambisana yiZhejiang University nePeking University. Yakhelwe phezu kohlaka lwe-OpenClaw Agent, futhi ixhumanisa lonke uchungechunge lokuphathwa kwezokuxhumana nge-ejenti eyodwa ehlakaniphile, kusukela ekutholeni izihloko ezidumile, ekuhleleni izihloko, eku...20
ChesoI-ejenti ye-AII-Cheso iyi-Agent yezethulo ezalwa yi-AI eyethulwe yi-Tencent. Ithuluzi lingenza ucwaningo ngokuzimele; umsebenzisi udinga kuphela ukuchaza izidingo ngolimi olujwayelekile, bese i-Cheso izosesha ulwazi ku-inthanethi ngokuzenzakalelayo, iqinisekise imithombo ngokuyiqhathanisa, ihlele ukwakheka kokuqukethwe, bese...10
zgI-ejenti ye-AIzg (zvec-grep) iyithuluzi elihlanganisiwe lokusesha elivulekile le-Zvec AI, eligxile ekusebenzeni endaweni futhi elakhelwe onjiniyela nama-AI Agent. Leli thuluzi lihlanganisa ukufanisa okunembile kwe-ripgrep, ukusesha kwamagama kwe-BM25 kanye nokusesha kwe-vector semantic kube yinto eyodwa, li...10
Lovart AII-ejenti ye-AII-Lovart AI iyi-ejenti ye-AI egxile enqubweni yokuklama, ethatha imisebenzi yokudala umbhalo, izithombe, amavidiyo nokumodela kwe-3D ngengxoxo. Ngemva kokuba umsebenzisi efake izidingo ngolimi olujwayelekile, uhlelo luhlela izinyathelo ezifanele, lukhiqize amaphosta, amavidiyo amafushane okukhangisa nokuqukethwe okubukwayo komkhiqizo, futhi luvumele ukuqhubeka nokukhetha nokulungisa, ukunweba isithombe noma ukulungisa imininingwane ethile kukhanvasi ehlakaniphile. Lufanele amaqembu aqalayo, abasebenzi bezokukhangisa noku-opharetha, abaklami nabadali abazimele, ukuze bahlolisise ngokushesha izinkomba ezibukwayo, benze izinto zokukhangisa futhi bahlele izinhlelo zomkhiqizo. Imiphumela ekhiqiziwe isadinga ukuhlolwa nokulungiswa ngumuntu; amaphrojekthi ayinkimbinkimbi akufanele athembele ngokuphelele enqubweni ezenzakalelayo. Ngokwemininingwane etholakalayo, umkhiqizo utholakala mahhala.009.1
ManusI-ejenti ye-AII-Manus ingumenzeli we-AI wemisebenzi eyinkimbinkimbi. Ayigcini ngokunikeza iziphakamiso zengxoxo, kodwa yehlukanisa izinyathelo ngokwezinjongo zomsebenzisi bese yenza imisebenzi efana nokuqoqa ulwazi, ukuhlaziya okuqukethwe nokucubungula ikhodi. Umsebenzisi uqala ngokuchaza umsebenzi, bese uhlelo luhlela, luphequlule amawebhusayithi, lufunde imithombo, luhlele imiphumela futhi lukhiqize umbiko, itafula noma okunye okungalethwa. Ifanele abacwaningi nabahlaziyi, onjiniyela, abaphathi bemikhiqizo kanye nabasebenzisi abadinga ukuphatha izindaba ezifana nokuhlela uhambo noma ukuhlunga izindlu. Ulwazi lukhombisa ukuthi i-Manus isekela ukusebenza okungavumelanisiwe nokusingathwa kwemisebenzi efwini, kodwa amakhono atholakalayo, ubude besikhathi somsebenzi kanye nekhwalithi yemiphumela kungase kuthintwe yinkimbinkimbi yomsebenzi nohlelo lwesevisi. Imiphumela ebalulekile kusafanele ihlolwe umuntu.009.1
MiniMax AgentI-ejenti ye-AII-MiniMax Agent ingumkhiqizo we-ejenti yewebhu ohlinzekwa yi-MiniMax, eyenzelwe abasebenzisi abadinga ukuphatha imisebenzi yansuku zonke, ucwaningo nemisebenzi ephindaphindwayo. Abasebenzisi bangadala imisebenzi ekhasini lewebhu, basebenzise ama-plugin, baxhumanise nomakhalekhukhwini, futhi bahlele ukuhamba komsebenzi ngamaqembu ama-Agent; ikhasi liphinde linikeze isixhumanisi sokulanda inguqulo yedeskithophu kanye nesixhumanisi sokubuka amanani ezinhlelo.068.4
PAST-BenchI-ejenti ye-AII-PAST-Bench iwuhlolo lokulinganisa olwakhiwe yithimba likaMengdi Wang eNyuvesi yasePrinceton, olusetshenziselwa ukuhlola ikhono le-Personal AI Agent lokuzithuthukisa ngokuphindaphindiwe. I-PAST-Bench iqhathanisa ukusebenza kwemisebenzi ngaphansi kwezimo ezinememori nezingenayo, ukuze inqume ukuthi ulwazi olugcinwe yi-Agent phakathi kwamaseshini luthuthukisa ngempela yini ukuziphatha okulandelayo...00
Nemotron 3.5 LightningI-ejenti ye-AII-Nemotron 3.5 Lightning iyimodeli ye-MoE yomthombo ovulekile enamapharamitha angu-30B eyethulwe ngu-NVIDIA, elungiselelwe ngokukhethekile amasistimu anama-agent amaningi. Uma iqhathaniswa namamodeli afanayo, ijubane lokukhipha imiphumela likhuphuka izikhathi ezi-4, kanti ijubane lokuqedela imisebenzi yama-agent likhuphuka ngo-30%, ...00
MindMemOS – Uhlelo lokusebenza lwememori lwe-AI Agent oluvulekele umphakathi, oluvulwe yi-Huawei Noah's Ark LabI-ejenti ye-AIMindMemOS – Uhlelo lokusebenza lwememori lwe-AI Agent oluvulekele umphakathi, oluvulwe yi-Huawei Noah's Ark Lab, lwenzelwe ukuhlinzeka ama-agent adinga ukuphathwa kwememori yesikhathi eside ngongqimba lwememori oluvulelekile nolungasetshenziswa kabusha. Lusebenzisa isakhiwo esinezilinganiso ezintathu: ibhizinisi, isibaluli nesikhathi, luhlukanisa imemori ku-Agent eyodwa luyenze impahla ezimele esebenza kuwo wonke amafreyimwokhi, futhi luthuthukisa imemori ngokuqhubekayo ngezinqubo ezifana ne-Dreaming yokuhlela ungaxhunyiwe ku-inthanethi, i-Feedback yokulungisa amaphutha kanye ne-Skill Evolution yokuziphendukela ngokuzenzakalelayo. Uhlelo luhlinzeka ngezindlela ezimbili zokubhala imemori (MindVanilla, MindSchema) kanye nokusesha kwe-Compact Search okunemizila eminingi, futhi lusekela ukufakwa ku-Docker endaweni, i-API esebenza efwini, i-Python SDK, i-CLI nezinye izindlela zokuxhuma. Ngokunemba kokubuyisa okuhlelekile okungaphezu kuka-94% kanye nokuthuthukisa ukunemba kwezimpendulo zemibuzo ngo-10% ngemva kokucindezelwa kwememori, lusiza abathuthukisi ukuthi babelane ngolwazi phakathi kwezingxoxo futhi bathuthukise ukusebenza kwemisebenzi ezimweni ezifana nabasizi abenziwe ngezifiso, ukubambisana kwama-Agent amaningi kanye ne-automation eyinkimbinkimbi yemisebenzi yasehhovisi.00
I-Kitesurf – isiphequluli esilula sefu se-AI Agent esethulwe yi-CloudflareI-ejenti ye-AII-Kitesurf isiphequluli esilula sefu se-AI Agent esethulwe yi-Cloudflare, esakhiwe nge-Rust futhi esisebenza endaweni ehlukanisiwe ye-Workers V8. Sisusa imisebenzi ebhekiswe kubantu, njengamathebhu nezandiso, sigxile ekuhlaziyweni kwe-DOM, ekusebenzeni kwe-JavaScript nasekukhipheni idatha. Siphinde sihambisane ne-CDP ne-MCP, kanti i-Puppeteer ne-Playwright kungaxhunywa kuso ngaphandle kokubhalwa kabusha. Uma siqhathaniswa ne-Chromium, sisebenzisa i-CPU engaphansi ngokuphindwe izikhathi ezi-3, imemori engaphansi ngokuphindwe izikhathi ezi-4.7 kuya kwezi-7, futhi sifanele imisebenzi yama-AI Agent enokuncintisana okuphezulu nengenazo izimo ezigciniwe. Abathuthukisi bangasebenzisa isixhumi esibonakalayo se-Quick Actions ukuze bathathe isithombe-skrini, bakhiphe i-HTML noma benze i-PDF ngokuchofoza okukodwa, noma bazame isevisi ku-inthanethi nge-Playground.00
Muse Glimmer – Imodeli yolimi enkulu enamapharamitha ayizigidi eziyizinkulungwane ezingu-30 evulwe yi-MetaI-ejenti ye-AIIMuse Glimmer imodeli yolimi enkulu enamapharamitha ayizigidi eziyizinkulungwane ezingu-30 evulwe yi-Meta, eyenzelwe futhi yalungiselelwa ngokujulile ukuqhuba ama-Workflow e-Agent ahlala kudivayisi ubusuku nemini. Ngobuchwepheshe be-4bit quantization, imodeli ingasebenza kahle ku-GPU eyikhadi elilodwa elinememori engu-24GB noma ku-Apple Silicon (M4/M5 Max) enememori ehlanganisiwe engu-32GB, cishe ngaphandle kokuncipha kokusebenza. Ihlanganisa i-DFlash speculative decoding, efinyelela ekusheshiseni okufika ku-3.1x ku-hardware esezingeni eliphezulu njenge-RTX 5090, futhi isekela imisebenzi ye-multimodal efana nezingxoxo zesikhathi sangempela, ukubiza amathuluzi, ukukhiqiza ikhodi kanye nokucabanga ngezinyathelo eziningi. Ifanele abathuthukisi ngabanye, amabhizinisi amancane naphakathi, kanye nezimboni ezifana nezempilo nezezimali ezinezidingo eziphezulu zobumfihlo bedatha. Inikeza amandla e-AI asebenza kahle endaweni ngaphandle kwemfuneko enkulu yememori ye-GPU, futhi yonke idatha ihlala ngaphakathi kwedivayisi.00
ClawTeamsI-ejenti ye-AII-ClawTeams iyinkundla yokubambisana ekuphathweni kwe-e-commerce eqhutshwa yi-AI. Abasebenzisi bathumela umgomo ngomusho owodwa ku-IM efana ne-Slack, Teams noma i-WeChat, bese i-AI Team Lead ihlukanisa imisebenzi ngokuzenzakalelayo, ihlele ama-Agent ochwepheshe angaphezu kuka-30 ukuthi ayenze kanyekanye, ...00
Grok BotI-ejenti ye-AII-Grok Bot iwuhlelo lwe-AI Agent olwenziwe yi-SpaceXAI, olukwazi ukungena kuma-akhawunti abasebenzisi futhi lusebenzise amawebhusayithi namathuluzi njengomuntu wangempela, lusebenzele abasebenzisi efwini amahora angu-24 ngosuku. Lolu hlelo lusekela ukusebenzisana kwama-Bot amaningi ngesikhathi esisodwa, ukurekhodwa...00
EvoXI-ejenti ye-AII-EvoX iyi-AI Agent ezithuthukisa yona eyethulwe yi-EvoMap. Isebenzisa imodi yeqoqo ukuhlukanisa imisebenzi eyinkimbinkimbi ibe yimisebenzi emincane enikezwa ama-Agent amaningi ukuthi ayenze ngesikhathi esisodwa, bese ihlanganisa imiphumela, igweme ukungcoliswa komongo. Ama-Agent angazuza kokuhlangenwe nakho komhlaba wonke ukuze azithuthukise,具...00
Orchard – Uhlaka lwe-Agentic AI oluvulelekile olwakhiwe yi-Microsoft ResearchI-ejenti ye-AII-Orchard – uhlaka lwe-Agentic AI oluvulelekile olwakhiwe yi-Microsoft Research. Uhlaka lugxile kusevisi yezindawo i-Orchard Env esekelwe ku-Kubernetes, futhi lusekela ukusetshenziswa kabusha kwama-sandbox ezizindeni ezahlukene ukuze kwenziwe i-data distillation, i-reinforcement learning rollout nokuhlola. Luhlanganisa izimo ezintathu ezinkulu: ubunjiniyela besoftware, ukuzulazula ku-browser kanye nomsizi womuntu siqu. Lunikeza ama-recipe aphelele okuqeqesha e-SFT + RL, kanye nama-trajectory angu-107K e-SWE nedatha engu-3,070 ye-GUI multimodal rollout, futhi lungakwazi ukuqeqeshwa ngokuqondile kusukela ekuqaleni kuye ekugcineni kuma-harness angempela afana ne-Codex ne-OpenClaw. Lubhekiswe emaqenjini ezemfundo nezimboni acwaninga futhi athuthukisa ama-Agent, i-Orchard inezindleko eziphansi zekhompyutha nezokusingathwa, futhi ikhono layo lokusebenza kuma-harness ahlukahlukene lisiza amamodeli amancane ukusondela ekusebenzeni kwezinhlelo ezihamba phambili.00
wigoloI-ejenti ye-AII-wigolo iyithuluzi lokusesha lendawo elinomthombo ovulekile, elixhuma kuma-Agent okuhlela afana ne-Claude Code ne-Cursor nge-MCP. Ayidingi i-API Key, futhi ngaphandle kwezindleko ungenza usesho, uthwebule amakhasi ewebhu, ukhase isayithi lonke, ukhiphe idatha ehlelekile, ugcine inqolobane futhi wenze ucwaningo endaweni yangakini...00
I-Wan有无界I-ejenti ye-AII-Wan有无界 – indawo yokusebenzisana yama-Agent anemisebenzi eminingi eyethulwe yi-Aliyun, eyenzelwe amabhizinisi nosomabhizinisi abazimele. Ihlanganisa ngokuzenzakalelayo abasebenzi bedijithali abafana ne-PMO, abathuthukisi nabahloli, bakhe iqembu lokusebenzisana eliqhubekisela phambili futhi lethule amaphrojekthi ayinkimbinkimbi ngokulandelana. Inkundla inikeza izifanekiso zezimo ezilungele ukusetshenziswa, ukuxhumeka okuguquguqukayo kwama-Agent asekhaya, kanye nokugcinwa nokusetshenziswa kabusha kwempahla ngokuzenzakalelayo. Isekela ukushintsha phakathi kwemodi yezingxoxo zohambo olulodwa kanye nemodi yengxoxo yeqembu enama-Agent amaningi. Iqukethe i-Agent yokuxhumanisa eyakhelwe ngaphakathi eqinisekisa ukuthi ukusebenzisana kwezindima eziningi akuphazamiseki, futhi yonke inqubo ibonakala futhi ilandeleka, okusiza abasebenzisi baguqule inqubo yokwenza ibe yimpahla yeqembu ne-SOP engasetshenziswa kabusha.00
I-Warp Agent CLI – ithuluzi elizimele lokubhala ikhodi ngomugqa womyalo elivela kwa-WarpI-ejenti ye-AII-Warp Agent CLI iyithuluzi elizimele lokubhala ikhodi ngomugqa womyalo elivela kwa-Warp, elisebenza kunoma iyiphi indawo yetheminali, okuhlanganisa i-Ghostty, i-iTerm 2, i-VS Code kanye netheminali eyakhelwe ohlelweni. Leli thuluzi lakhiwe phezu kwengqalasizinda yetheminali ye-Warp, linemultiplexer eyakhelwe ngaphakathi ephethe ukuxhumana kwe-pty ngokwemvelo, futhi ligcina umongo we-Agent lapho ushintsha phakathi kwezinkomba noma kumaseshini e-SSH. Lingasebenzisa izinhlelo ezisebenzisanayo zesikrini esigcwele ezifana ne-sqlite ne-python, lisekele iphrothokholi ye-MCP, futhi lihlele ngokuzenzakalelayo imisebenzi yama-Agent angaphansi asebenza endaweni noma efwini, okuhlanganisa ukuyinikezela ku-Claude Code noma ku-Codex.00
ForgeStencil – uhlelo lwe-Stencil lokucwaninga nokuthunyelwa okuzenzakalelayo ngokuphelele oluqaliswe yi-Mianbi IntelligentI-ejenti ye-AII-ForgeStencil – uhlelo lwe-Stencil lokucwaninga nokuthunyelwa okuzenzakalelayo ngokuphelele oluqaliswe yi-Mianbi Intelligent, lwakhiwe ngokubambisana yi-Mianbi Intelligent nomphakathi we-OpenBMB. Lusebenzisa ukwakheka okuvaliwe kwama-agent amabili, i-Kernel Agent ne-App Agent. Lungenza ngokuzenzakalelayo yonke inqubo kusukela ekutholeni amasu okuthuthukisa, ekuhlanganiseni ama-CUDA Kernel asebenza kahle, kuya ekuhlanganiseni nasekuqinisekiseni izinhlelo ezisezingeni lokukhiqiza, ngaphandle kokungenelela komuntu. Esontweni elilodwa, uhlelo luthuthukise izinhlelo ezingaphezu kwe-100 zokubala zezimboni nezesayensi kusukela ekuqaleni kuye ekugcineni, lwafinyelela ekusheshiseni okumaphakathi okuphindwe ngo-1.41. Luhlanganisa imikhakha yezimboni engu-8 kanye nemikhakha yesayensi engu-5, okuhlanganisa ukuhlola uwoyela negesi, ukulingisa amandla kazibuthe kagesi, nezithombe zezokwelapha, ngaleyo ndlela lwehlise kakhulu ukuncika kochwepheshe be-HPC nesikhathi sokuthuthukisa.00
Today AII-ejenti ye-AII-Today AI iwuhlelo lokusebenza lwe-AI Agent oluqondene nomuntu siqu, olwethulwe ngu-Qi Junyuan, owayengumsunguli we-Teambition. Umkhiqizo ungaxhuma kukhompyutha, amafayela akuleyo khompyutha, imilayezo, ikhalenda kanye nezinsizakalo ezihlukahlukene zamafu, uqhubeke nokwakha inkumbulo yesikhathi eside yomsebenzisi, ufeze ukusuka...00
Qwen-CUA – I-Computer Use Agent yomdabu eyethulwe yi-Alibaba Qwen nabanyeI-ejenti ye-AII-Qwen-CUA – I-Computer Use Agent yomdabu eyethulwe yi-Alibaba Qwen nabanye, isekelwe ekwakhiweni kwe-Mixture-of-Experts engu-397B-A17B. Ibona isimo sesixhumi esibonakalayo ngezithombe zesikrini kuphela, ngaphandle kokuncika esihlahleni se-DOM noma kumethadatha yokufinyeleleka, futhi ikhiphe ngokuqondile imicimbi yekhibhodi negundane ukuze ilawule iziphequluli, amadeskithophu nezinhlelo zobuchwepheshe. Ekuhlolweni kwe-OSWorld-Verified ithole amaphuzu angu-86.2, kanti inguqulo enkulu i-Qwen-CUA-Max ithole angu-87.6, okuyizinga elihamba phambili emkhakheni wama-CUA omthombo ovulekile. Isebenzisa ukuqeqeshwa kokuphela-kuya-ekupheleni kokubona-isenzo, inokusebenziseka ezinkundleni ezahlukene nokuzinza okuphezulu, futhi sekukhishwe umbiko wezobuchwepheshe kanye ne-Demo, okuyenza ifanele ukuhlolwa okuzenzakalelayo, i-RPA, usizo lokufinyeleleka kanye nokulawula izinhlelo zocwaningo.00
MemHarness – uhlaka lokwakha kabusha inkumbulo oluqalwe yi-Shanghai AI Lab nabanyeI-ejenti ye-AII-MemHarness iwuhlaka lokwakha kabusha inkumbulo lwama-LLM Agent, olwethulwe ngokuhlanganyela yi-Shanghai AI Lab namanyuvesi ahlanganisa i-Zhejiang University, i-Fudan University kanye ne-Shanghai Jiao Tong University. Ifaka izinyathelo zokuhlaziya nokwakha kabusha phakathi kokusesha inkumbulo nokwenza isenzo, igcine, ibhale kabusha noma ilahle okuhlangenwe nakho kwangaphambilini ngokuhambisana nomongo wamanje. Isebenzisa ukuqeqeshwa kwe-GRPO kusukela ekuqaleni kuye ekugcineni ngaphandle kwedatha eyengeziwe enelebula eyenziwe abantu. Uhlaka lusekela yonke inqubo yokusesha inkumbulo, ukuhlaziya nokwakha kabusha, ukukhiqiza izenzo, ukubuyisela okuhlangenwe nakho nokususa inkumbulo enenani eliphansi, futhi lunikeza inqubo yokuthatha izinqumo enezigaba ezinhlanu. Imodeli enamapharamitha angu-7B idlula i-Gemini-2.5-Pro ngo-23.1% ku-ALFWorld nango-39.7% ku-WebShop, kanti izinga layo lokuphumelela elijwayelekile ezimweni ezingaphandle kokusatshalaliswa lifinyelela ku-85.9%. Isebenziseka emisebenzini yasekhaya yamarobhothi ahlakaniphile, ekuthengeni ku-inthanethi ngokuzenzakalelayo, ekusekelweni kwamakhasimende okuhlakaniphile, ekusizeni ukuthuthukiswa kwekhodi nakwezinye izimo.10Wonke amathuluzi akulesi sigaba ayaboniswa.