Ṣiṣe awọn aṣoju lori awọn awoṣe aala ti iṣakoso loni-tabi iyalo agbara GPU, gbe awọn iwuwo tirẹ lọ, ati ipa awọn iṣẹ ṣiṣe Digio si awọn aaye ipari ikọkọ ni aaye iṣẹ kanna.
Fi awoṣe aiyipada fun aṣoju tabi danu fun iṣẹ-ṣiṣe kan. Lilo jẹ mita ni Digio Tokens lati iwọntunwọnsi ero rẹ — apamọwọ kanna boya aṣoju n pe Sonnet, GPT-4o, tabi 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.
Alakoso le ṣeduro Sonnet vs Opus vs awoṣe filasi din owo ti o da lori iru iṣẹ-ṣiṣe. Awọn olumulo agbara ṣeto awọn aseku fun ipa aṣoju-iwadi lori Sonnet, atunyẹwo ikẹhin lori Opus, fifi aami le lori Haiku tabi 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.
Ṣe o nilo atunṣe-itanran, aaye ayẹwo-afẹfẹ, tabi idiyele idiyele asọtẹlẹ bi? Ṣafikun agbara GPU igbẹhin si aaye iṣẹ Digio rẹ, fi sori ẹrọ akopọ iṣẹ ti o fẹ, ati awọn aṣoju tọka si aaye ipari ikọkọ rẹ.
Awọn apa GPU wakati tabi oṣooṣu (A100, H100, L40S kilasi) ti a so mọ agbatọju rẹ — ya sọtọ si awọn alabara miiran.
Po si safetensors, GGUF, tabi fa lati rẹ iforukọsilẹ; ṣiṣe Llama, Mistral, Qwen, ati awọn ohun orin ipe ti aṣa.
vLLM, TGI, Ollama, tabi awọn aworan apoti ti o ṣetọju-Awọn aṣoju Digio pe OpenAI-ibaramu URL URL kan.
Lati ṣe, iwiregbe ẹgbẹ, awọn ọgbọn, ati ifowosowopo ko yipada — nikan ni ifẹhinti ifọkansi jẹ tirẹ.
Firanṣẹ awọn igbesẹ ifura si GPU aladani ati lo Claude tabi GPT fun iwadii gbogbo eniyan ni ṣiṣan iṣẹ kan.
Iwoye VPC, egress aimi, awọn akọọlẹ iṣayẹwo, ati awọn atokọ iyọọda awoṣe fun awọn ẹgbẹ ti a ṣe ilana.
Iṣeto deede lati odo si awọn aṣoju ti n pe aaye ipari rẹ:
Yan VRAM, agbegbe, ati akoko iṣẹ (ti nwaye la nigbagbogbo-lori). Ibi ipamọ fun awọn ọkọ oju omi iwuwo pẹlu apẹẹrẹ tabi gbe garawa rẹ soke.
Bẹrẹ aworan iṣẹ tabi SSH sinu, fi awọn awakọ CUDA sori ẹrọ, ati fifuye awọn aaye ayẹwo. Awọn sọwedowo ilera jẹrisi awoṣe ti ṣetan.
Ṣafikun URL ipilẹ, bọtini API, ati id awoṣe ni awọn eto aaye iṣẹ. Digio fọwọsi lairi ati ọna kika ami ṣaaju ki o to lọ laaye.
Mu awoṣe ikọkọ rẹ bi aiyipada fun awọn aṣoju ti a yan; Awọn awoṣe Claude / GPT ti iṣakoso wa ni ẹgbẹ ni ẹgbẹ.
Yiyalo GPU jẹ idiyele lọtọ lati awọn ṣiṣe alabapin ero Digio. Kan si wa fun igbero agbara, SLAs, ati ijira lati inu iṣupọ itọkasi ti o wa.
Yiyan awọn API ti a ṣakoso pẹlu itọka ti a gbalejo lori Digio.
Ṣiṣe alabapin Digio rẹ ni wiwa awọn amayederun, awọn aṣoju, ati pẹlu Digio Tokens. Awọn debiti lilo awoṣe ti iṣakoso ti iwọntunwọnsi ami nipasẹ awọn ami titẹ sii/jade gangan. Yiyalo GPU jẹ afikun fun awọn ẹrọ ti o ṣakoso.
Bẹẹni-aṣoju kọọkan le ni aiyipada tirẹ. Awọn iṣẹ-ṣiṣe ati awọn ibaraẹnisọrọ le bori fun ṣiṣe ẹyọkan laisi iyipada aiyipada agbaye.
Opus ti wa ni aifwy fun ero lile ati awọn ero isomọ gigun; Sonnet yiyara ati din owo fun awọn losiwajulosehin aṣoju lojoojumọ. Haiku ati awọn awoṣe kilasi filasi dara julọ fun awọn iṣẹ-ṣiṣe iwọn didun.
Awọn aaye iṣẹ ile-iṣẹ le ni ihamọ awọn olupese awoṣe ti njade ati ipa ọna gbogbo ijabọ aṣoju si aaye ipari GPU rẹ. Ipo arabara jẹ aiyipada fun ọpọlọpọ awọn ẹgbẹ.
Awọn ẹbun da lori agbegbe ati ibeere — ni igbagbogbo 24–80 GB VRAM awọn ipele fun awọn awoṣe kilasi 7B–70B ati awọn apa GPU pupọ fun awọn akopọ nla. A ṣe iranlọwọ iwọn VRAM lati iṣiro paramita rẹ ati titobi.
Orchestration (awọn aṣoju, awọn iṣẹ-ṣiṣe, ibi ipamọ) duro lori ero rẹ. Itọkasi lori GPU rẹ jẹ idiyele bi akoko GPU; o le ni iyan mita lilo apẹrẹ aami fun gbigba agbara inu.
Awọn aṣoju, awọn iṣẹ-ṣiṣe, ati awọn ọgbọn-aaye iṣẹ ni ayika awoṣe ti o yan.
Bẹrẹ lori Claude ati GPT loni, lẹhinna ṣafikun GPU igbẹhin nigbati o ba ṣetan lati gbalejo awọn iwuwo aṣa-awọn aṣoju kanna, awọn iṣẹ ṣiṣe kanna, itọkasi rẹ.
