--- title: KAT-Coder-V2.5-Dev vs Qwen3.6-35B: Is 8x Fewer Tokens... description: A hands-on comparison of KAT-Coder-V2.5-Dev (a fine-tuned version of Qwen3.6-35B by Quai Pilot) against the base Qwen3.6-35B model for agentic coding tasks.... canonical: https://daily.dev/posts/kat-coder-v2-5-dev-vs-qwen3-6-35b-is-8x-fewer-tokens-actually-better-for-agentic-coding--jqvo5mh3e twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: KAT-Coder-V2.5-Dev vs Qwen3.6-35B: Is 8x Fewer Tokens Actually Better for Agentic Coding? | daily.dev og:description: A hands-on comparison of KAT-Coder-V2.5-Dev (a fine-tuned version of Qwen3.6-35B by Quai Pilot) against the base Qwen3.6-35B model for agentic coding tasks.... og:url: https://daily.dev/posts/kat-coder-v2-5-dev-vs-qwen3-6-35b-is-8x-fewer-tokens-actually-better-for-agentic-coding--jqvo5mh3e og:image: https://api.daily.dev/og/posts/JQVO5mH3E.png og:image:alt: KAT-Coder-V2.5-Dev vs Qwen3.6-35B: Is 8x Fewer Tokens Actually Better for Agentic Coding? 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. # KAT-Coder-V2.5-Dev vs Qwen3.6-35B: Is 8x Fewer Tokens Actually Better for Agentic Coding? **[Execute Automation](https://daily.dev/sources/executeautomation)** · 15 min read · 0 upvotes · 0 comments ## Summary A hands-on comparison of KAT-Coder-V2.5-Dev (a fine-tuned version of Qwen3.6-35B by Quai Pilot) against the base Qwen3.6-35B model for agentic coding tasks. The fine-tuned model uses SFT on 127,000 examples plus reinforcement learning to dramatically reduce token usage and tool calls. In a side-by-side test using a Python coding agent on local hardware (via OMLX with 4-bit quantization), KAT-Coder generated only 183K tokens vs 1.4M for Qwen — 7x fewer — with 9.7x less cache reads and 3.4x fewer assistant turns. The key insight is that the base Qwen model re-reads its own plan before almost every file, causing context bloat and RL training instability, while the fine-tuned model reads once and moves on. Benchmark results on SWE-bench verified also show KAT-Coder outperforming the base model on agentic coding tasks. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: --- Tags: [#llm](https://daily.dev/tags/llm), [#deep-learning](https://daily.dev/tags/deep-learning), [#ai-coding](https://daily.dev/tags/ai-coding), [#reinforcement-learning](https://daily.dev/tags/reinforcement-learning), [#local-ai](https://daily.dev/tags/local-ai) [View this post on daily.dev](https://daily.dev/posts/kat-coder-v2-5-dev-vs-qwen3-6-35b-is-8x-fewer-tokens-actually-better-for-agentic-coding--jqvo5mh3e) ```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":"KAT-Coder-V2.5-Dev vs Qwen3.6-35B: Is 8x Fewer Tokens Actually Better for Agentic Coding?","url":"https://daily.dev/posts/kat-coder-v2-5-dev-vs-qwen3-6-35b-is-8x-fewer-tokens-actually-better-for-agentic-coding--jqvo5mh3e","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/kat-coder-v2-5-dev-vs-qwen3-6-35b-is-8x-fewer-tokens-actually-better-for-agentic-coding--jqvo5mh3e"},"datePublished":"2026-08-05T09:17:31.010Z","dateModified":"2026-08-05T09:17:52.461Z","description":"A hands-on comparison of KAT-Coder-V2.5-Dev (a fine-tuned version of Qwen3.6-35B by Quai Pilot) against the base Qwen3.6-35B model for agentic coding tasks....","image":"https://i.ytimg.com/vi/ox6upguli7U/sddefault.jpg","thumbnailUrl":"https://i.ytimg.com/vi/ox6upguli7U/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/kat-coder-v2-5-dev-vs-qwen3-6-35b-is-8x-fewer-tokens-actually-better-for-agentic-coding--jqvo5mh3e","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"llm,deep-learning,ai-coding,reinforcement-learning,local-ai","timeRequired":"PT15M","video":{"@type":"VideoObject","name":"KAT-Coder-V2.5-Dev vs Qwen3.6-35B: Is 8x Fewer Tokens Actually Better for Agentic Coding?","description":"A hands-on comparison of KAT-Coder-V2.5-Dev (a fine-tuned version of Qwen3.6-35B by Quai Pilot) against the base Qwen3.6-35B model for agentic coding tasks....","thumbnailUrl":"https://i.ytimg.com/vi/ox6upguli7U/sddefault.jpg","uploadDate":"2026-08-05T09:17:31.010Z","duration":"PT15M","url":"https://api.daily.dev/r/JQVO5mH3E","embedUrl":"https://www.youtube.com/embed/ox6upguli7U"}} {"@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":"KAT-Coder-V2.5-Dev vs Qwen3.6-35B: Is 8x Fewer Tokens Actually Better for Agentic Coding?"}]} ```