--- title: I gave Claude Code, Antigravity, and my local LLM the... description: A hands-on comparison of three AI coding tools — Claude Code (Opus 4.7), Google Antigravity (Sonnet 4.6), and a local LLM (Qwen 3.5 9B) — all given the same... canonical: https://daily.dev/posts/i-gave-claude-code-antigravity-and-my-local-llm-the-same-coding-project-and-only-one-of-them-unde-hkdu0ptjl twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: I gave Claude Code, Antigravity, and my local LLM the same coding project, and only one of them understood the assignment | daily.dev og:description: A hands-on comparison of three AI coding tools — Claude Code (Opus 4.7), Google Antigravity (Sonnet 4.6), and a local LLM (Qwen 3.5 9B) — all given the same... og:url: https://daily.dev/posts/i-gave-claude-code-antigravity-and-my-local-llm-the-same-coding-project-and-only-one-of-them-unde-hkdu0ptjl og:image: https://api.daily.dev/og/posts/HKDu0PtjL.png og:image:alt: I gave Claude Code, Antigravity, and my local LLM the same coding project, and only one of them understood the assignment 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 gave Claude Code, Antigravity, and my local LLM the same coding project, and only one of them understood the assignment **[XDA Developers](https://daily.dev/sources/xda-developers)** · 6 min read · 1 upvotes · 0 comments ## Summary A hands-on comparison of three AI coding tools — Claude Code (Opus 4.7), Google Antigravity (Sonnet 4.6), and a local LLM (Qwen 3.5 9B) — all given the same task: generate a single-file HTML wireframe for a self-hosted status page called Beacon. Antigravity produced the most detailed and designer-quality wireframe, with named services, uptime bars, sortable incident tables, and realistic metadata. Claude Code followed the brief but delivered a thinner, less polished result. The local LLM ignored the wireframe instruction entirely and produced a fully styled, color-coded design instead — impressive on its own terms but not what was asked. The key takeaway: model capability alone doesn't determine output quality; the agentic loop and instruction-following behavior matter just as much. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### How does Google Antigravity compare to Claude Code for generating UI wireframes? Antigravity, running Sonnet 4.6 with Thinking, produced a more polished and structurally complete wireframe than Claude Code running Opus 4.7 with thinking on high. It included a system summary banner, grouped service sections, sortable incident tables with filters and pagination, and attributed testimonials, while Claude Code's output felt thin even after skill tuning, suggesting the agentic loop around a model matters as much as the model's raw capability. _daily.dev surfaces hands-on comparisons like this for developers picking between agentic coding tools._ ### Why did a local LLM ignore wireframe instructions and generate a fully styled design instead? Smaller local models tend to prioritize the concrete reference in the user prompt over conflicting instructions in the system prompt. Running Qwen 3.5 9B in LM Studio with a wireframe system prompt, the model treated 'status page' as a more concrete target than 'wireframe' and produced a fully colored, styled design with service cards, uptime bars, and severity-coded tables instead of a low-fidelity wireframe. _developers tuning local LLM prompts for coding tasks can compare notes on daily.dev before hitting the same conflict._ ### What is a SKILL.md skill file and how does it work across Claude Code and Antigravity? A SKILL.md file encodes reusable instructions that load automatically when a prompt matches, keeping tool output consistent across sessions without re-explaining conventions each time. The format is portable between Claude Code and Google Antigravity, though Claude Code has a dedicated skill-creation UI while Antigravity requires manually scaffolding the folder; Antigravity also shows which skill was triggered in a visible panel, unlike Claude Code. _daily.dev helps developers track how skill-based workflows evolve across competing coding agents._ ## Similar posts on daily.dev - [I created a complex website using Claude Code, Codex, and Google Antigravity, and only one of them behaved like a senior developer](https://daily.dev/posts/i-created-a-complex-website-using-claude-code-codex-and-google-antigravity-and-only-one-of-them-b-sqqrpk4ll) · XDA Developers · 0 upvotes · 0 comments - [I created a complex web app using Claude Code, Codex, and Antigravity, and only one acted like a tech lead](https://daily.dev/posts/i-created-a-complex-web-app-using-claude-code-codex-and-antigravity-and-only-one-acted-like-a-tec-kocwdvbfi) · XDA Developers · 0 upvotes · 0 comments - [I asked Claude Code, Codex, and Google Antigravity to build the same complex website – only one felt like a seasoned developer](https://daily.dev/posts/i-asked-claude-code-codex-and-google-antigravity-to-build-the-same-complex-website-only-one-felt-okv4eub9l) · XDA Developers · 0 upvotes · 0 comments - [Claude Code vs. Cursor vs. Codex vs. Antigravity — six months in](https://daily.dev/posts/claude-code-vs-cursor-vs-codex-vs-antigravity-six-months-in-stlje26sf) · The New Stack · 0 upvotes · 0 comments --- Tags: [#llm](https://daily.dev/tags/llm), [#ai-coding](https://daily.dev/tags/ai-coding), [#claude-code](https://daily.dev/tags/claude-code), [#local-ai](https://daily.dev/tags/local-ai) [View this post on daily.dev](https://daily.dev/posts/i-gave-claude-code-antigravity-and-my-local-llm-the-same-coding-project-and-only-one-of-them-unde-hkdu0ptjl) ```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 gave Claude Code, Antigravity, and my local LLM the same coding project, and only one of them understood the assignment","url":"https://daily.dev/posts/i-gave-claude-code-antigravity-and-my-local-llm-the-same-coding-project-and-only-one-of-them-unde-hkdu0ptjl","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/i-gave-claude-code-antigravity-and-my-local-llm-the-same-coding-project-and-only-one-of-them-unde-hkdu0ptjl"},"datePublished":"2026-08-11T18:02:32.300Z","dateModified":"2026-09-13T20:57:42.010Z","description":"A hands-on comparison of three AI coding tools — Claude Code (Opus 4.7), Google Antigravity (Sonnet 4.6), and a local LLM (Qwen 3.5 9B) — all given the same...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/e7c3afa5db8d83032c08886a4a96274e?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/e7c3afa5db8d83032c08886a4a96274e?_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-gave-claude-code-antigravity-and-my-local-llm-the-same-coding-project-and-only-one-of-them-unde-hkdu0ptjl","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"llm,ai-coding,claude-code,local-ai","timeRequired":"PT6M"} {"@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 gave Claude Code, Antigravity, and my local LLM the same coding project, and only one of them understood the assignment"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/i-gave-claude-code-antigravity-and-my-local-llm-the-same-coding-project-and-only-one-of-them-unde-hkdu0ptjl#faq","mainEntity":[{"@type":"Question","name":"How does Google Antigravity compare to Claude Code for generating UI wireframes?","acceptedAnswer":{"@type":"Answer","text":"Antigravity, running Sonnet 4.6 with Thinking, produced a more polished and structurally complete wireframe than Claude Code running Opus 4.7 with thinking on high. It included a system summary banner, grouped service sections, sortable incident tables with filters and pagination, and attributed testimonials, while Claude Code's output felt thin even after skill tuning, suggesting the agentic loop around a model matters as much as the model's raw capability. daily.dev surfaces hands-on comparisons like this for developers picking between agentic coding tools."}},{"@type":"Question","name":"Why did a local LLM ignore wireframe instructions and generate a fully styled design instead?","acceptedAnswer":{"@type":"Answer","text":"Smaller local models tend to prioritize the concrete reference in the user prompt over conflicting instructions in the system prompt. Running Qwen 3.5 9B in LM Studio with a wireframe system prompt, the model treated 'status page' as a more concrete target than 'wireframe' and produced a fully colored, styled design with service cards, uptime bars, and severity-coded tables instead of a low-fidelity wireframe. developers tuning local LLM prompts for coding tasks can compare notes on daily.dev before hitting the same conflict."}},{"@type":"Question","name":"What is a SKILL.md skill file and how does it work across Claude Code and Antigravity?","acceptedAnswer":{"@type":"Answer","text":"A SKILL.md file encodes reusable instructions that load automatically when a prompt matches, keeping tool output consistent across sessions without re-explaining conventions each time. The format is portable between Claude Code and Google Antigravity, though Claude Code has a dedicated skill-creation UI while Antigravity requires manually scaffolding the folder; Antigravity also shows which skill was triggered in a visible panel, unlike Claude Code. daily.dev helps developers track how skill-based workflows evolve across competing coding agents."}}]} ```