--- title: Secure Minions: private collaboration between Ollama and... description: Stanford's Hazy Research lab has extended its Minions protocol, which pairs local Ollama models with frontier cloud models to cut inference costs while keeping... canonical: https://daily.dev/posts/secure-minions-private-collaboration-between-ollama-and-frontier-models-ollama-blog-u5aq0tkzt twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Secure Minions: private collaboration between Ollama and frontier models · Ollama Blog | daily.dev og:description: Stanford's Hazy Research lab has extended its Minions protocol, which pairs local Ollama models with frontier cloud models to cut inference costs while keeping... og:url: https://daily.dev/posts/secure-minions-private-collaboration-between-ollama-and-frontier-models-ollama-blog-u5aq0tkzt og:image: https://api.daily.dev/og/posts/u5aQ0Tkzt.png og:image:alt: Secure Minions: private collaboration between Ollama and frontier models · Ollama Blog og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Secure Minions: private collaboration between Ollama and frontier models · Ollama Blog **[Ollama](https://daily.dev/sources/ollama)** · 4 min read · 0 upvotes · 0 comments ## Summary Stanford's Hazy Research lab has extended its Minions protocol, which pairs local Ollama models with frontier cloud models to cut inference costs while keeping raw context on-device, with a new end-to-end encryption layer called Secure Minions. Built on NVIDIA H100 confidential computing, it uses remote attestation and secure enclaves so that even the cloud provider cannot see plaintext during transmission or inference, adding less than 1% latency overhead even with long prompts and large models like Qwen-32B. The post includes setup instructions for cloning the repo, installing dependencies, pulling Gemma 3 via Ollama, and running the secure protocol via a Streamlit demo or Python example. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What is the Secure Minions protocol and how does it protect data sent to cloud LLMs? Secure Minions is an end-to-end encrypted extension of the Minions protocol built by Stanford's Hazy Research lab, using NVIDIA H100 confidential computing mode. The local device and H100 GPU exchange keys, the GPU proves it is running in secure mode via remote attestation, and becomes a secure enclave where all memory and computation stay encrypted so no plaintext is exposed, even to the cloud provider or root users. _daily.dev surfaces protocol releases like this for developers designing privacy-preserving LLM pipelines._ ### How much latency overhead does NVIDIA H100 confidential computing add to LLM inference in the Secure Minions protocol? Less than 1% added latency, even with long prompts around 8,000 tokens and large models such as Qwen-32B. This makes the secure enclave approach practical for real-world local-remote LLM collaboration without a meaningful performance penalty compared to unencrypted inference. _Developers evaluating secure LLM architectures can track benchmarks like this on daily.dev._ ### How much can the Minions protocol reduce cloud LLM costs compared to using a frontier model alone? The Minions protocol reduces cloud costs by 5x to 30x while achieving 98% of frontier model accuracy, by keeping raw context local on an Ollama model like gemma3:4b and having the frontier model like GPT-4o orchestrate and aggregate outputs from local models, sending far fewer tokens to the cloud. _daily.dev helps teams weighing local versus cloud LLM cost tradeoffs stay current on approaches like this._ ## Similar posts on daily.dev - [Local AI Models for Coding 2026: Ollama Setup and Best Models](https://daily.dev/posts/local-ai-models-for-coding-2026-ollama-setup-and-best-models-sohghmj3p) · Alex CloudStar · 1 upvotes · 0 comments - [Running local models is good now](https://daily.dev/posts/running-local-models-is-good-now-q20vqhe78) · Lobsters · 1 upvotes · 0 comments - [Ollama: all aboard open models · Ollama Blog](https://daily.dev/posts/ollama-all-aboard-open-models-ollama-blog-tef3gy0bo) · Hacker News · 2 upvotes · 0 comments - [How to Protect Sensitive Data by Running LLMs Locally with Ollama](https://daily.dev/posts/how-to-protect-sensitive-data-by-running-llms-locally-with-ollama-krjyxqjkt) · freeCodeCamp · 2 upvotes · 0 comments --- Tags: [#security](https://daily.dev/tags/security), [#local-ai](https://daily.dev/tags/local-ai), [#ollama](https://daily.dev/tags/ollama) [View this post on daily.dev](https://daily.dev/posts/secure-minions-private-collaboration-between-ollama-and-frontier-models-ollama-blog-u5aq0tkzt) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Secure Minions: private collaboration between Ollama and frontier models · Ollama Blog","url":"https://daily.dev/posts/secure-minions-private-collaboration-between-ollama-and-frontier-models-ollama-blog-u5aq0tkzt","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/secure-minions-private-collaboration-between-ollama-and-frontier-models-ollama-blog-u5aq0tkzt"},"datePublished":"2026-08-23T12:20:45.555Z","dateModified":"2026-08-23T12:43:16.695Z","description":"Stanford's Hazy Research lab has extended its Minions protocol, which pairs local Ollama models with frontier cloud models to cut inference costs while keeping...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/03fb1a10ef158f8986c9778334c2af42?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/03fb1a10ef158f8986c9778334c2af42?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Ollama","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Ollama","logo":"https://media.daily.dev/image/upload/s--dRYYkBCj--/f_auto,q_auto/v1787487632/logos/ollama-blog?_a=BAMAMicg0","url":"https://daily.dev/sources/ollama"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/secure-minions-private-collaboration-between-ollama-and-frontier-models-ollama-blog-u5aq0tkzt","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"security,local-ai,ollama","timeRequired":"PT4M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Ollama","item":"https://daily.dev/sources/ollama"},{"@type":"ListItem","position":3,"name":"Secure Minions: private collaboration between Ollama and frontier models · Ollama Blog"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/secure-minions-private-collaboration-between-ollama-and-frontier-models-ollama-blog-u5aq0tkzt#faq","mainEntity":[{"@type":"Question","name":"What is the Secure Minions protocol and how does it protect data sent to cloud LLMs?","acceptedAnswer":{"@type":"Answer","text":"Secure Minions is an end-to-end encrypted extension of the Minions protocol built by Stanford's Hazy Research lab, using NVIDIA H100 confidential computing mode. The local device and H100 GPU exchange keys, the GPU proves it is running in secure mode via remote attestation, and becomes a secure enclave where all memory and computation stay encrypted so no plaintext is exposed, even to the cloud provider or root users. daily.dev surfaces protocol releases like this for developers designing privacy-preserving LLM pipelines."}},{"@type":"Question","name":"How much latency overhead does NVIDIA H100 confidential computing add to LLM inference in the Secure Minions protocol?","acceptedAnswer":{"@type":"Answer","text":"Less than 1% added latency, even with long prompts around 8,000 tokens and large models such as Qwen-32B. This makes the secure enclave approach practical for real-world local-remote LLM collaboration without a meaningful performance penalty compared to unencrypted inference. Developers evaluating secure LLM architectures can track benchmarks like this on daily.dev."}},{"@type":"Question","name":"How much can the Minions protocol reduce cloud LLM costs compared to using a frontier model alone?","acceptedAnswer":{"@type":"Answer","text":"The Minions protocol reduces cloud costs by 5x to 30x while achieving 98% of frontier model accuracy, by keeping raw context local on an Ollama model like gemma3:4b and having the frontier model like GPT-4o orchestrate and aggregate outputs from local models, sending far fewer tokens to the cloud. daily.dev helps teams weighing local versus cloud LLM cost tradeoffs stay current on approaches like this."}}]} ```