--- title: My local LLM can call every tool that Claude can, except... description: Running a local LLM with full tool-calling capabilities comparable to Claude is achievable on consumer hardware. Using the Pi agent harness with MCP and... canonical: https://daily.dev/posts/my-local-llm-can-call-every-tool-that-claude-can-except-it-runs-on-my-own-hardware-du52dfxnm twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: My local LLM can call every tool that Claude can, except it runs on my own hardware | daily.dev og:description: Running a local LLM with full tool-calling capabilities comparable to Claude is achievable on consumer hardware. Using the Pi agent harness with MCP and... og:url: https://daily.dev/posts/my-local-llm-can-call-every-tool-that-claude-can-except-it-runs-on-my-own-hardware-du52dfxnm og:image: https://api.daily.dev/og/posts/du52DFXNM.png og:image:alt: My local LLM can call every tool that Claude can, except it runs on my own hardware 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. # My local LLM can call every tool that Claude can, except it runs on my own hardware **[XDA Developers](https://daily.dev/sources/xda-developers)** · 7 min read · 2 upvotes · 0 comments ## Summary Running a local LLM with full tool-calling capabilities comparable to Claude is achievable on consumer hardware. Using the Pi agent harness with MCP and pi-subagents extensions, the author connects Playwright for browser automation, SearXNG for web search, Crawl4AI for content extraction, and Qdrant/mem0 for persistent memory — all self-hosted. The setup runs on an AMD RX 7900 XTX (Qwen3.6 27B at 40-60 tokens/sec) and a Lenovo ThinkStation with 128GB unified memory (Qwen3.6 35B MoE). Key insight: tool-calling reliability matters more than parameter count, and modern quantized models like Qwen3.6 retain their tool-use capabilities at Q4/Q8 quants without meaningful degradation. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### Can a local LLM do tool calling like Claude Code with browser automation, web search, and memory? Yes, a local setup can replicate Claude's tool categories using the Pi agent harness with the MCP extension connected to Playwright for browser automation, SearXNG for web search, Crawl4AI for scraping, and Qdrant plus mem0 for persistent memory. Models like Qwen3.6-27B or 35B-A3B run this stack on consumer GPUs such as an RX 7900 XTX with 24GB VRAM. _Developers piecing together a local agent stack can compare tool-calling setups shared on daily.dev._ ### Does quantizing a model to Q4 hurt its tool-calling ability? No, tool-calling ability typically holds up across Q4 and Q8 quantization without meaningful degradation in models trained with tool use in mind, such as Qwen3.6. A 27-billion-parameter model quantized to Q4_K_XL fits in 24GB of VRAM and still reliably picks the right tool and arguments during agent workflows. _Anyone tuning local model quantization for agent workflows can track findings like this on daily.dev._ ### What is the Pi agent harness and how does it differ from Claude Code? Pi is a minimal, extensible agent harness that ships with only four base tools (Read, Write, Edit, Bash), unlike Claude Code's sealed, fully built-out system. Users extend Pi themselves with extensions like MCP for external tool access and pi-subagents for spawning worker processes, giving more control over building agent capabilities on their own terms. _Developers weighing agent harness options can follow comparisons like this on daily.dev._ ## Similar posts on daily.dev - [The biggest local LLM on your machine is useless if it can't call a single tool, no matter how many parameters it has](https://daily.dev/posts/the-biggest-local-llm-on-your-machine-is-useless-if-it-can-t-call-a-single-tool-no-matter-how-many--uauiyyggz) · XDA Developers · 1 upvotes · 0 comments - [Can a Local LLM Run My AI Assistant?](https://daily.dev/posts/can-a-local-llm-run-my-ai-assistant--k8pjpamza) · Towards Data Science · 0 upvotes · 0 comments - [I used a local Qwen3.6-35B-A3B LLM to extend my coding harness' capabilities](https://daily.dev/posts/i-used-a-local-qwen3-6-35b-a3b-llm-to-extend-my-coding-harness-capabilities-nthq4whz5) · XDA Developers · 0 upvotes · 0 comments - [I ran Qwen3.5 locally instead of Claude Code. Here’s what happened.](https://daily.dev/posts/i-ran-qwen3-5-locally-instead-of-claude-code-here-s-what-happened--fnmwj9ey5) · InfoWorld · 0 upvotes · 0 comments --- Tags: [#llm](https://daily.dev/tags/llm), [#ai-agents](https://daily.dev/tags/ai-agents), [#mcp](https://daily.dev/tags/mcp), [#local-ai](https://daily.dev/tags/local-ai), [#qwen](https://daily.dev/tags/qwen) [View this post on daily.dev](https://daily.dev/posts/my-local-llm-can-call-every-tool-that-claude-can-except-it-runs-on-my-own-hardware-du52dfxnm) ```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":"My local LLM can call every tool that Claude can, except it runs on my own hardware","url":"https://daily.dev/posts/my-local-llm-can-call-every-tool-that-claude-can-except-it-runs-on-my-own-hardware-du52dfxnm","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/my-local-llm-can-call-every-tool-that-claude-can-except-it-runs-on-my-own-hardware-du52dfxnm"},"datePublished":"2026-07-07T12:31:10.362Z","dateModified":"2026-09-14T08:51:16.534Z","description":"Running a local LLM with full tool-calling capabilities comparable to Claude is achievable on consumer hardware. Using the Pi agent harness with MCP and...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/ee6a9b0f32ab3f9a0c8ebf9ce518ba90?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/ee6a9b0f32ab3f9a0c8ebf9ce518ba90?_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/my-local-llm-can-call-every-tool-that-claude-can-except-it-runs-on-my-own-hardware-du52dfxnm","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":2},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"llm,ai-agents,mcp,local-ai,qwen","timeRequired":"PT7M"} {"@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":"My local LLM can call every tool that Claude can, except it runs on my own hardware"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/my-local-llm-can-call-every-tool-that-claude-can-except-it-runs-on-my-own-hardware-du52dfxnm#faq","mainEntity":[{"@type":"Question","name":"Can a local LLM do tool calling like Claude Code with browser automation, web search, and memory?","acceptedAnswer":{"@type":"Answer","text":"Yes, a local setup can replicate Claude's tool categories using the Pi agent harness with the MCP extension connected to Playwright for browser automation, SearXNG for web search, Crawl4AI for scraping, and Qdrant plus mem0 for persistent memory. Models like Qwen3.6-27B or 35B-A3B run this stack on consumer GPUs such as an RX 7900 XTX with 24GB VRAM. Developers piecing together a local agent stack can compare tool-calling setups shared on daily.dev."}},{"@type":"Question","name":"Does quantizing a model to Q4 hurt its tool-calling ability?","acceptedAnswer":{"@type":"Answer","text":"No, tool-calling ability typically holds up across Q4 and Q8 quantization without meaningful degradation in models trained with tool use in mind, such as Qwen3.6. A 27-billion-parameter model quantized to Q4_K_XL fits in 24GB of VRAM and still reliably picks the right tool and arguments during agent workflows. Anyone tuning local model quantization for agent workflows can track findings like this on daily.dev."}},{"@type":"Question","name":"What is the Pi agent harness and how does it differ from Claude Code?","acceptedAnswer":{"@type":"Answer","text":"Pi is a minimal, extensible agent harness that ships with only four base tools (Read, Write, Edit, Bash), unlike Claude Code's sealed, fully built-out system. Users extend Pi themselves with extensions like MCP for external tool access and pi-subagents for spawning worker processes, giving more control over building agent capabilities on their own terms. Developers weighing agent harness options can follow comparisons like this on daily.dev."}}]} ```