Klique deepe s its collabo atio with NVIDIA th ough platfo m suppo t fo NVIDIA Nemot o 3.5 Supe VL, Ope Shell, Dy amo, a d NeMo Switchya d.
SAN FRANCISCO, CA / ACCESS Newswi e / Octobe 21, 2026 / Klique, a e te p ise co t ol pla e fo AI esou ce ma ageme t, gove a ce, a d cost visibility, today a ou ced a suite of ew i teg atio s with NVIDIA desig ed to optimize GPU pe fo ma ce a d delive e ha ced capabilities fo e te p ises.
Buildi g o a st o g fou datio of its collabo atio with NVIDIA, Klique helpi g e te p ises pai NVIDIA’s cutti g-edge i f ast uctu e with secu e e te p ise executio ac oss th ee key ope atio al a eas:
Optimizi g Compute Cost a d Data Sove eig tyThe NVIDIA Nemot o family of ope models p ovides ope , high-efficie cy, low-late cy pe fo ma ce desig ed fo high-volume age tic wo kflows o a si gle GPU. By u i g Nemot o 3.5 Supe VL o Klique, e te p ise use s ca deploy NVIDIA’s model di ectly o thei ow ha dwa e, e su i g o data leaves the te a t while elimi ati g thi d-pa ty pe -toke billi g. Compleme ti g this, Klique custome s ca ow leve age NVIDIA NeMo Switchya d’s i tellige t model outi g algo ithms alo gside Klique’s LLM gateway, givi g o ga izatio s the flexibility to optimize spe d o f o tie models o e su e st ict data sove eig ty.
Ru i g Age ts at P oductio ScaleA si gle age t task ca t igge ma y model calls as the age t pla s, uses tools, a d easo s th ough each step, so i fe e ce dema d climbs fast o ce age ts each p oductio . NVIDIA Dy amo is built fo that load: it dist ibutes i fe e ce ac oss GPU wo ke s a d se ds epeated equests to the wo ke s that al eady hold the eleva t cached co text. Klique custome s ca deploy NVIDIA Dy amo di ectly f om the Klique platfo m a d u age ts at p oductio scale o thei ow GPU i f ast uctu e o i the cloud, with Klique addi g both the secu ity access ights a d obse vability as well as LLM outi g maki g su e la ge scale age tic wo kflows a e dive ted i to cost effective AI esou ces. Teams ca also use Klique to et ai age ts o the esults of thei past wo k, a d the platfo m eco ds the data, model ve sio , a d compute cost behi d eve y update so each cha ge ca be eviewed a d t aced. With Dy amo se vi g i fe e ce ac oss thei GPUs a d Klique t acki g eve y age t’s usage, cost, a d cha ges, e te p ises ca scale age ts to p oductio dema d o ha dwa e they al eady ow .
Secu i g Auto omous Age t Wo kflowsNVIDIA Ope Shell establishes a secu e, mo ito ed u time e vi o me t desig ed to p otect i te al files, etwo ks, a d c ede tials du i g age t executio . By i teg ati g Ope Shell i to the platfo m, Klique allows e te p ises to deploy auto omous age ts like those built with NemoClaw ope bluep i ts, Claude Code, Pi, He mes, a d Codex with total co fide ce, p ovidi g the gove a ce laye , e te p ise policy complia ce, a d audit t ails equi ed fo p oductio age tic wo kflows.
“Scali g p oductio -g ade age tic wo kflows should ‘t fo ce e te p ises to choose betwee spi ali g cloud costs a d data p ivacy,” said Moses Guttma , CEO a d CO-fou de , Klique. “Ou expa ded collabo atio with NVIDIA pai s low-late cy models like Nemot o 3.5 Supe VL a d sma t outi g algo ithms via NeMo Switchya d with Klique’s cost co t ol a d gove a ce f amewo k. This allows ou custome s to maximize GPU utilizatio , mai tai st ict data sove eig ty, a d u f o tie -level AI wo kloads di ectly o thei i f ast uctu e without thi d-pa ty pe -toke i flatio .”
“Eve y e te p ise is wo ki g out how to u age ts efficie tly a d u de its ow policies,” said Joh Fa elli, vice p eside t, E te p ise Softwa e at NVIDIA. “By buildi g o NVIDIA Nemot o , Dy amo a d Ope Shell, Klique helps custome s put age ts to wo k withi thei u ique e vi o me ts.”
To lea mo e about Klique’s i teg atio s with NVIDIA, visit https://klique.ai.
About Klique
Klique (fo me ly Clea ML) is the co t ol pla e fo e te p ise AI, whe e the people buildi g AI, the people usi g it, a d the IT teams u i g the i f ast uctu e all wo k u de the same ules. It outes eve y AI equest to the ight comme cial o self-hosted model based o cost, complexity, a d data se sitivity; e fo ces budgets, quotas, a d access policies o ce, eve ywhe e; a d ma ages GPUs a d compute o a y silico , f om Kube etes a d ba e metal to fully ai -gapped e vi o me ts. Eve y toke , GPU cycle, a d dolla is t aced back to the p oject that spe t it, so AI spe d g ows o ly as fast as policy allows. Headqua te ed i Be keley, Califo ia, Klique is t usted by mo e tha 2,100 o ga izatio s a d 300,000 AI builde s wo ldwide. Lea mo e at klique.ai.
Co tact I fo matio
Re ce Holt Pho e: +14048196709
SOURCE: Klique
View the o igi al p ess elease o ACCESS Newswi e
