--- title: I turned my old phone into a local LLM server, and it... description: An old Poco M6 Pro phone was repurposed into a local LLM inference server using llama.cpp compiled via Termux. After installing dependencies and compiling with... canonical: https://daily.dev/posts/i-turned-my-old-phone-into-a-local-llm-server-and-it-handles-productivity-tasks-better-than-i-expec-lo5simets twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: I turned my old phone into a local LLM server, and it handles productivity tasks better than I expected | daily.dev og:description: An old Poco M6 Pro phone was repurposed into a local LLM inference server using llama.cpp compiled via Termux. After installing dependencies and compiling with... og:url: https://daily.dev/posts/i-turned-my-old-phone-into-a-local-llm-server-and-it-handles-productivity-tasks-better-than-i-expec-lo5simets og:image: https://api.daily.dev/og/posts/lo5SIMeTS.png og:image:alt: I turned my old phone into a local LLM server, and it handles productivity tasks better than I expected 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. # I turned my old phone into a local LLM server, and it handles productivity tasks better than I expected **[XDA Developers](https://daily.dev/sources/xda-developers)** · 4 min read · 0 upvotes · 0 comments ## Summary An old Poco M6 Pro phone was repurposed into a local LLM inference server using llama.cpp compiled via Termux. After installing dependencies and compiling with parallel jobs to speed up the process, Gemma 4 E2B was chosen for its Per-Layer Embeddings architecture, which gives it reasoning capabilities similar to larger 5.1B models despite a 2.3B-sized footprint. The setup handles OCR on documents (using vision capabilities), PDF and note summarization, and email proofreading while away from home. For heavier workloads like coding or complex automation, dedicated home lab nodes running larger models like Qwen3.6-35B-A3B remain necessary. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### Can I run a local LLM on an old Android phone using llama.cpp? Yes, llama.cpp can be compiled directly on Android through Termux and used to run small models like Gemma 4 E2B. Install dependencies with apt install git cmake libandroid-spawn, clone the llama.cpp repository, then compile with cmake -B build followed by cmake --build build --config Release -j 8 to speed up compilation using parallel jobs on limited hardware. _daily.dev surfaces setups like this for developers exploring low-power local LLM hosting._ ### Why is Gemma 4 E2B good for running on low-power devices like an old phone? Gemma 4 E2B uses Per-Layer Embeddings, letting each decoding layer use its own embedding vector per token, which speeds up lookups without hurting computation quality. This gives it reasoning capabilities comparable to a 5.1B model while keeping a footprint similar to a 2.3B model, making it well suited for OCR, document summarization, and email proofreading on constrained hardware. _Developers comparing small local models for edge devices track findings like these via daily.dev._ ## Similar posts on daily.dev - [I turned my phone into a local LLM server, and it handles vision, voice, and tool calls](https://daily.dev/posts/i-turned-my-phone-into-a-local-llm-server-and-it-handles-vision-voice-and-tool-calls-f2fcaxaha) · XDA Developers · 0 upvotes · 0 comments --- Tags: [#self-hosting](https://daily.dev/tags/self-hosting), [#local-ai](https://daily.dev/tags/local-ai), [#gemma](https://daily.dev/tags/gemma), [#llama-cpp](https://daily.dev/tags/llama-cpp) [View this post on daily.dev](https://daily.dev/posts/i-turned-my-old-phone-into-a-local-llm-server-and-it-handles-productivity-tasks-better-than-i-expec-lo5simets) ```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":"I turned my old phone into a local LLM server, and it handles productivity tasks better than I expected","url":"https://daily.dev/posts/i-turned-my-old-phone-into-a-local-llm-server-and-it-handles-productivity-tasks-better-than-i-expec-lo5simets","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/i-turned-my-old-phone-into-a-local-llm-server-and-it-handles-productivity-tasks-better-than-i-expec-lo5simets"},"datePublished":"2026-09-01T00:01:52.910Z","dateModified":"2026-09-01T00:18:38.025Z","description":"An old Poco M6 Pro phone was repurposed into a local LLM inference server using llama.cpp compiled via Termux. After installing dependencies and compiling with...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/4d5b723a1ada3b93f0698dd16bf4db31?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/4d5b723a1ada3b93f0698dd16bf4db31?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"XDA Developers","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":"XDA Developers","logo":"https://media.daily.dev/image/upload/s--kCg6yyAP--/f_auto,q_auto/v1774964407/logos/xda-developers?_a=BAMAMiWQ0","url":"https://daily.dev/sources/xda-developers"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/i-turned-my-old-phone-into-a-local-llm-server-and-it-handles-productivity-tasks-better-than-i-expec-lo5simets","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"self-hosting,local-ai,gemma,llama-cpp","timeRequired":"PT4M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"XDA Developers","item":"https://daily.dev/sources/xda-developers"},{"@type":"ListItem","position":3,"name":"I turned my old phone into a local LLM server, and it handles productivity tasks better than I expected"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/i-turned-my-old-phone-into-a-local-llm-server-and-it-handles-productivity-tasks-better-than-i-expec-lo5simets#faq","mainEntity":[{"@type":"Question","name":"Can I run a local LLM on an old Android phone using llama.cpp?","acceptedAnswer":{"@type":"Answer","text":"Yes, llama.cpp can be compiled directly on Android through Termux and used to run small models like Gemma 4 E2B. Install dependencies with apt install git cmake libandroid-spawn, clone the llama.cpp repository, then compile with cmake -B build followed by cmake --build build --config Release -j 8 to speed up compilation using parallel jobs on limited hardware. daily.dev surfaces setups like this for developers exploring low-power local LLM hosting."}},{"@type":"Question","name":"Why is Gemma 4 E2B good for running on low-power devices like an old phone?","acceptedAnswer":{"@type":"Answer","text":"Gemma 4 E2B uses Per-Layer Embeddings, letting each decoding layer use its own embedding vector per token, which speeds up lookups without hurting computation quality. This gives it reasoning capabilities comparable to a 5.1B model while keeping a footprint similar to a 2.3B model, making it well suited for OCR, document summarization, and email proofreading on constrained hardware. Developers comparing small local models for edge devices track findings like these via daily.dev."}}]} ```