Gudanar da wakilai akan samfuran kan iyakokin da aka sarrafa a yau-ko yin hayan ƙarfin GPU, ƙaddamar da ma'aunin ku, da hanyar ayyukan Digio zuwa wuraren ƙarewa masu zaman kansu a cikin filin aiki iri ɗaya.
Sanya tsoho samfurin kowane wakili ko soke kowane ɗawainiya. An ƙididdige amfani a cikin Digio Tokens daga ma'auni na shirinku - walat iri ɗaya ko wakilin ya kira Sonnet, GPT-4o, ko Gemini Flash.
Model list and token economics evolve with provider releases. Your workspace shows live options when you assign a model to an agent; Digio Tokens debit from the same balance as in pricing.
Mai Gudanarwa na iya ba da shawarar Sonnet vs Opus vs samfurin filasha mai rahusa dangane da nau'in ɗawainiya. Masu amfani da wutar lantarki sun saita gazawar kowane rawar wakili-bincike akan Sonnet, bita na ƙarshe akan Opus, babban alamar alama akan Haiku ko Gemini Flash.
Per agent — default model in agent settings; override in To do or chat when needed.
Metered fairly — input, output, and cached tokens map to Digio Token charges (see usage in your wallet).
Skills stay the same — tools and integrations work across models; only latency and cost profile change.
Plan limits — more agents and monthly Digio Tokens on higher tiers; top up anytime on the pricing page.
Kuna buƙatar gyara mai kyau, wurin bincike mai tazara, ko farashin ƙima? Ƙara ƙarfin GPU da aka keɓe zuwa filin aikin ku na Digio, shigar da tarin hidimar da kuka fi so, da nuna wakilai a ƙarshen ƙarshen ku na sirri.
Ƙididdigar GPU na sa'a ko wata-wata (A100, H100, L40S class) wanda aka haɗe zuwa mai haya-wanda ke keɓe daga sauran abokan ciniki.
Loda safetensors, GGUF, ko ja daga wurin yin rajista; gudanar da Llama, Mistral, Qwen, da nagartattun waƙoƙi na al'ada.
vLLM, TGI, Ollama, ko hotunan kwantena da kuke kiyayewa-Ma'aikatan Digio suna kiran URL tushe mai jituwa na OpenAI.
Don yin, taɗi ta ƙungiya, ƙwarewa, da haɗin gwiwa ba su canza ba - kawai abin da ake so shine naku.
Aika matakai masu mahimmanci zuwa GPU masu zaman kansu kuma amfani da Claude ko GPT don bincike na jama'a a cikin aiki ɗaya.
VPC peering, a tsaye egress, duba rajistan ayyukan, da samfurin izini ga ƙungiyoyin da aka tsara.
Saitin al'ada daga sifili zuwa wakilai masu kiran ƙarshen ƙarshenku:
Zaɓi VRAM, yanki, da lokacin aiki (fashe vs koyaushe-kan). Adana don ma'aunin nauyi tare da misalin ko hawa guga.
Fara hoton hidima ko SSH a ciki, shigar da direbobin CUDA, da loda wuraren bincike. Binciken lafiya ya tabbatar da samfurin ya shirya.
Ƙara URL na tushe, maɓallin API, da ƙirar ƙira a cikin saitunan sararin aiki. Digio yana tabbatar da latency da tsarin alama kafin tafiya kai tsaye.
Zaɓi samfurin ku na sirri azaman tsoho don zaɓaɓɓun wakilai; Samfurin Claude/GPT da aka sarrafa ya kasance yana nan gefe da gefe.
Ana biyan hayar GPU daban daga biyan kuɗin shirin Digio. Tuntube mu don tsara iyawa, SLAs, da ƙaura daga gungu mai mahimmanci.
Zaɓin APIs da aka sarrafa vs ba da shawarar kai tsaye akan Digio.
Biyan kuɗin Digio ɗin ku ya ƙunshi abubuwan more rayuwa, wakilai, da haɗa Digio Tokens. Samfuran da aka sarrafa na amfani da ƙira yana ba da ma'auni na alamar ta ainihin shigarwa/fitarwa token. Hayan GPU ƙari ne don injinan da kuke sarrafawa.
Ee-kowane wakili na iya samun nasa tsoho. Ayyuka da taɗi na iya ƙetare don gudu ɗaya ba tare da canza tsohowar duniya ba.
An kunna Opus don ƙarin tunani da tsare-tsare masu tsayi; Sonnet yana da sauri kuma mai rahusa don madaukai wakili na yau da kullun. Haiku da nau'ikan filashi-flash sun fi dacewa don ƙananan ayyukan ƙara.
Wuraren aiki na kasuwanci na iya ƙuntata masu samarwa da ke waje da kuma tafiyar da duk zirga-zirgar wakili zuwa ƙarshen GPU ɗin ku. Yanayin haɗin gwiwa shine tsoho don yawancin ƙungiyoyi.
Abubuwan bayarwa sun dogara da yanki da buƙatu-yawanci 24–80 GB VRAM tiers don ƙirar aji 7B–70B da nodes-GPU da yawa don manyan tari. Muna taimakawa girman VRAM daga ƙidayar sigar ku da ƙididdigewa.
Orchestration (wakilai, ayyuka, ajiya) yana tsayawa akan shirin ku. Ana lissafin ƙididdigewa akan GPU ɗinku azaman lokacin GPU; Kuna iya ba da zaɓin amfani da sifar alama ta mita don dawo da cajin ciki.
Wakilai, ɗawainiya, da ƙwarewa — filin aiki a kusa da ƙirar da kuka zaɓa.
Fara a kan Claude da GPT a yau, sannan ƙara GPU sadaukarwa lokacin da kuke shirye don ɗaukar nauyin ma'aunin al'ada - wakilai iri ɗaya, ayyuka iri ɗaya, ra'ayin ku.
