--- title: Why New Gemma 4 QAT Model Fails in Local Coding Agent... description: A hands-on comparison of Google's Gemma 4 26B QAT (8-bit quantization-aware training) model versus Qwen 3.6 35B for local coding agent tasks inside VS Code.... canonical: https://daily.dev/posts/why-new-gemma-4-qat-model-fails-in-local-coding-agent-tasks-qwen-3-6-wins-owuo3msd0 twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Why New Gemma 4 QAT Model Fails in Local Coding Agent Tasks — QWEN 3.6 Wins | daily.dev og:description: A hands-on comparison of Google's Gemma 4 26B QAT (8-bit quantization-aware training) model versus Qwen 3.6 35B for local coding agent tasks inside VS Code.... og:url: https://daily.dev/posts/why-new-gemma-4-qat-model-fails-in-local-coding-agent-tasks-qwen-3-6-wins-owuo3msd0 og:image: https://api.daily.dev/og/posts/OwUo3msD0.png og:image:alt: Why New Gemma 4 QAT Model Fails in Local Coding Agent Tasks — QWEN 3.6 Wins 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. # Why New Gemma 4 QAT Model Fails in Local Coding Agent Tasks — QWEN 3.6 Wins **[Execute Automation](https://daily.dev/sources/executeautomation)** · 14 min read · 1 upvotes · 0 comments ## Summary A hands-on comparison of Google's Gemma 4 26B QAT (8-bit quantization-aware training) model versus Qwen 3.6 35B for local coding agent tasks inside VS Code. The test involves generating multi-turn conversational RAG test cases using the DeepEval library. Gemma 4 QAT produced syntactically incomplete and functionally incorrect test code, while Qwen 3.6 generated a well-structured, more complete pytest file including retrieval context building, session fixtures, and a judge model. Both models ran at roughly 53–54 tokens/second locally. The conclusion is that Qwen 3.6 significantly outperforms Gemma 4 QAT for coding tasks, despite Gemma's architectural improvements and memory efficiency claims. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Similar posts on daily.dev - [I replaced Gemma 4 with a two-year-old coding LLM, and my local AI setup works better than ever](https://daily.dev/posts/i-replaced-gemma-4-with-a-two-year-old-coding-llm-and-my-local-ai-setup-works-better-than-ever-gjv50xvwp) · XDA Developers · 0 upvotes · 0 comments - [I ran Gemma 4 and Qwen 3.5 for the same local tasks, and one pulled miles ahead](https://daily.dev/posts/i-ran-gemma-4-and-qwen-3-5-for-the-same-local-tasks-and-one-pulled-miles-ahead-ukuc7wocd) · XDA Developers · 0 upvotes · 0 comments --- Tags: [#data-science](https://daily.dev/tags/data-science), [#ai-assisted-development](https://daily.dev/tags/ai-assisted-development), [#local-ai](https://daily.dev/tags/local-ai), [#gemma](https://daily.dev/tags/gemma), [#qwen](https://daily.dev/tags/qwen) [View this post on daily.dev](https://daily.dev/posts/why-new-gemma-4-qat-model-fails-in-local-coding-agent-tasks-qwen-3-6-wins-owuo3msd0) ```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":"Why New Gemma 4 QAT Model Fails in Local Coding Agent Tasks — QWEN 3.6 Wins","url":"https://daily.dev/posts/why-new-gemma-4-qat-model-fails-in-local-coding-agent-tasks-qwen-3-6-wins-owuo3msd0","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/why-new-gemma-4-qat-model-fails-in-local-coding-agent-tasks-qwen-3-6-wins-owuo3msd0"},"datePublished":"2026-06-08T08:25:31.258Z","dateModified":"2026-06-10T16:19:55.416Z","description":"A hands-on comparison of Google's Gemma 4 26B QAT (8-bit quantization-aware training) model versus Qwen 3.6 35B for local coding agent tasks inside VS Code....","image":"https://i.ytimg.com/vi/cftzfRV5PhI/sddefault.jpg","thumbnailUrl":"https://i.ytimg.com/vi/cftzfRV5PhI/sddefault.jpg","isAccessibleForFree":true,"articleSection":"Execute Automation","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":"Execute Automation","logo":"https://media.daily.dev/image/upload/s--Uh-EE9Cf--/f_auto,q_auto/v1780213688/logos/executeautomation?_a=BAMAMiWQ0","url":"https://daily.dev/sources/executeautomation"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/why-new-gemma-4-qat-model-fails-in-local-coding-agent-tasks-qwen-3-6-wins-owuo3msd0","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"data-science,ai-assisted-development,local-ai,gemma,qwen","timeRequired":"PT14M","video":{"@type":"VideoObject","name":"Why New Gemma 4 QAT Model Fails in Local Coding Agent Tasks — QWEN 3.6 Wins","description":"A hands-on comparison of Google's Gemma 4 26B QAT (8-bit quantization-aware training) model versus Qwen 3.6 35B for local coding agent tasks inside VS Code....","thumbnailUrl":"https://i.ytimg.com/vi/cftzfRV5PhI/sddefault.jpg","uploadDate":"2026-06-08T08:25:31.258Z","duration":"PT14M","url":"https://api.daily.dev/r/OwUo3msD0","embedUrl":"https://www.youtube.com/embed/cftzfRV5PhI"}} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Execute Automation","item":"https://daily.dev/sources/executeautomation"},{"@type":"ListItem","position":3,"name":"Why New Gemma 4 QAT Model Fails in Local Coding Agent Tasks — QWEN 3.6 Wins"}]} ```