--- title: After trying 20+ local LLMs, these are the 4 I keep... description: A personal walkthrough of a curated local LLM setup running on KoboldCpp, with one model assigned per task: Qwen 3.6 27B for serious coding, Gemma 4 12B as an... canonical: https://daily.dev/posts/after-trying-20-local-llms-these-are-the-4-i-keep-installed-for-different-jobs-h3rljguwn twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: After trying 20+ local LLMs, these are the 4 I keep installed for different jobs | daily.dev og:description: A personal walkthrough of a curated local LLM setup running on KoboldCpp, with one model assigned per task: Qwen 3.6 27B for serious coding, Gemma 4 12B as an... og:url: https://daily.dev/posts/after-trying-20-local-llms-these-are-the-4-i-keep-installed-for-different-jobs-h3rljguwn og:image: https://api.daily.dev/og/posts/h3rljgUWn.png og:image:alt: After trying 20+ local LLMs, these are the 4 I keep installed for different jobs 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. # After trying 20+ local LLMs, these are the 4 I keep installed for different jobs **[XDA Developers](https://daily.dev/sources/xda-developers)** · 4 min read · 1 upvotes · 0 comments ## Summary A personal walkthrough of a curated local LLM setup running on KoboldCpp, with one model assigned per task: Qwen 3.6 27B for serious coding, Gemma 4 12B as an everyday all-rounder, GPT OSS 20B for deeper reasoning and problem-solving, and Llama 3.3 8B for creative writing. The author explains the trade-offs (size, speed, quality) that led to keeping each model instead of constantly testing new ones. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### Which local LLM is best for coding tasks when running models with KoboldCpp? Qwen 3.6 27B, specifically a quantized Q5_K_S build around 18GB, is used for serious coding work including code creation, debugging, refactoring, and understanding existing code. It takes longer to generate larger code but produces cleaner, more usable results, making it worth the wait when code quality matters more than speed. _Developers weighing local coding models against cloud options can find more comparisons on daily.dev._ ### What is a good lightweight local LLM for everyday quick tasks? Gemma 4 12B works well as an everyday all-rounder for quick questions, summarizing, explaining concepts, and brainstorming. Its smaller size loads faster and feels more responsive than larger models, while leaving more system resources free for other tasks, making it a convenient default choice for simple requests. _Anyone building a local AI toolkit can track model comparisons like this on daily.dev._ ### Which local model is better suited for complex reasoning tasks instead of quick answers? GPT OSS 20B is used for multi-layered questions that require breaking down problems, comparing options, and checking whether ideas hold up, since it spends more time working through reasoning rather than returning a fast response. The main trade-off is slower generation, especially with larger prompts. _daily.dev helps developers compare reasoning-focused local models before committing storage and compute._ ## Similar posts on daily.dev - [I tested 3 local LLMs on my actual work — and each model won at something different](https://daily.dev/posts/i-tested-3-local-llms-on-my-actual-work-and-each-model-won-at-something-different-d643xkwyx) · XDA Developers · 0 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 - [Best Local LLM Models 2026](https://daily.dev/posts/best-local-llm-models-2026-e56ssvqkw) · SitePoint · 2 upvotes · 0 comments - [I tested every tiny local LLM worth running, and only three survived the cut](https://daily.dev/posts/i-tested-every-tiny-local-llm-worth-running-and-only-three-survived-the-cut-ldlh8vwq9) · XDA Developers · 0 upvotes · 0 comments - [Running local models on an M4 with 24GB memory](https://daily.dev/posts/running-local-models-on-an-m4-with-24gb-memory-iezpeho6j) · Hacker News · 4 upvotes · 0 comments --- Tags: [#data-science](https://daily.dev/tags/data-science), [#llm](https://daily.dev/tags/llm), [#self-hosting](https://daily.dev/tags/self-hosting), [#local-ai](https://daily.dev/tags/local-ai) [View this post on daily.dev](https://daily.dev/posts/after-trying-20-local-llms-these-are-the-4-i-keep-installed-for-different-jobs-h3rljguwn) ```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":"After trying 20+ local LLMs, these are the 4 I keep installed for different jobs","url":"https://daily.dev/posts/after-trying-20-local-llms-these-are-the-4-i-keep-installed-for-different-jobs-h3rljguwn","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/after-trying-20-local-llms-these-are-the-4-i-keep-installed-for-different-jobs-h3rljguwn"},"datePublished":"2026-08-21T16:33:38.377Z","dateModified":"2026-09-13T20:34:54.990Z","description":"A personal walkthrough of a curated local LLM setup running on KoboldCpp, with one model assigned per task: Qwen 3.6 27B for serious coding, Gemma 4 12B as an...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/d3b8b59674df3d8c3107c1f9f6af3390?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/d3b8b59674df3d8c3107c1f9f6af3390?_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/after-trying-20-local-llms-these-are-the-4-i-keep-installed-for-different-jobs-h3rljguwn","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"data-science,llm,self-hosting,local-ai","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":"After trying 20+ local LLMs, these are the 4 I keep installed for different jobs"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/after-trying-20-local-llms-these-are-the-4-i-keep-installed-for-different-jobs-h3rljguwn#faq","mainEntity":[{"@type":"Question","name":"Which local LLM is best for coding tasks when running models with KoboldCpp?","acceptedAnswer":{"@type":"Answer","text":"Qwen 3.6 27B, specifically a quantized Q5_K_S build around 18GB, is used for serious coding work including code creation, debugging, refactoring, and understanding existing code. It takes longer to generate larger code but produces cleaner, more usable results, making it worth the wait when code quality matters more than speed. Developers weighing local coding models against cloud options can find more comparisons on daily.dev."}},{"@type":"Question","name":"What is a good lightweight local LLM for everyday quick tasks?","acceptedAnswer":{"@type":"Answer","text":"Gemma 4 12B works well as an everyday all-rounder for quick questions, summarizing, explaining concepts, and brainstorming. Its smaller size loads faster and feels more responsive than larger models, while leaving more system resources free for other tasks, making it a convenient default choice for simple requests. Anyone building a local AI toolkit can track model comparisons like this on daily.dev."}},{"@type":"Question","name":"Which local model is better suited for complex reasoning tasks instead of quick answers?","acceptedAnswer":{"@type":"Answer","text":"GPT OSS 20B is used for multi-layered questions that require breaking down problems, comparing options, and checking whether ideas hold up, since it spends more time working through reasoning rather than returning a fast response. The main trade-off is slower generation, especially with larger prompts. daily.dev helps developers compare reasoning-focused local models before committing storage and compute."}}]} ```