--- title: GitHub Copilot CLI gets a cross-model review agent... description: GitHub Copilot CLI has added an experimental feature called Rubber Duck that uses a second AI model from a different family to review the primary agent's work... canonical: https://daily.dev/posts/github-copilot-cli-gets-a-cross-model-review-agent-called-rubber-duck-bwobwp3qu twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: GitHub Copilot CLI gets a cross-model review agent called Rubber Duck | daily.dev og:description: GitHub Copilot CLI has added an experimental feature called Rubber Duck that uses a second AI model from a different family to review the primary agent's work... og:url: https://daily.dev/posts/github-copilot-cli-gets-a-cross-model-review-agent-called-rubber-duck-bwobwp3qu og:image: https://api.daily.dev/og/posts/BwObWP3QU.png og:image:alt: GitHub Copilot CLI gets a cross-model review agent called Rubber Duck 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. # GitHub Copilot CLI gets a cross-model review agent called Rubber Duck **[Collections](https://daily.dev/sources/collections)** · 2 min read · 1 upvotes · 0 comments ## Summary GitHub Copilot CLI has added an experimental feature called Rubber Duck that uses a second AI model from a different family to review the primary agent's work before execution. When Claude Sonnet 4.6 is the primary orchestrator, GPT-5.4 acts as the reviewer. The reviewer triggers automatically after plan generation, complex multi-file implementations, and test writing. On SWE-Bench Pro, this pairing closed 74.7% of the performance gap between Sonnet and Opus, with the biggest gains on complex multi-file problems. The core insight is that self-review is ineffective because a model that made a flawed assumption will likely defend it — a different model family provides genuinely independent scrutiny at Sonnet-level pricing. ## Content GitHub has added an experimental feature to Copilot CLI called Rubber Duck: a review agent that uses a model from a *different* AI family to critique the primary agent's plans before anything gets executed. The setup is straightforward. When Claude Sonnet 4.6 is the primary orchestrator, GPT-5.4 acts as the reviewer. The two models don't share a lineage, which is the whole point — a model reviewing its own output tends to miss the same things it missed the first time. Bringing in a different model family means genuinely independent scrutiny. ## How it works Rubber Duck triggers automatically at three checkpoints: - After the primary agent produces a plan - After complex multi-file implementations - After writing tests It can also be invoked manually. The reviewer's job is to look for missed details, questionable assumptions, and edge cases — not to rewrite anything, just to flag problems before they propagate. ## The benchmark numbers On SWE-Bench Pro, pairing Claude Sonnet 4.6 with Rubber Duck running GPT-5.4 closed 74.7% of the performance gap between Sonnet and Opus. The biggest gains showed up on complex multi-file problems, which makes sense — that's exactly where a flawed premise in step one becomes expensive by step ten. The cost argument is also worth noting: you get most of Opus-level performance at Sonnet-level pricing. ## Why this matters AI coding agents fail in a specific, annoying way. They don't usually produce obviously broken code. They produce polished, internally consistent output built on a wrong assumption. By the time you notice, the assumption has been baked into a dozen files. Self-review doesn't fix this. A model that made a flawed assumption in planning will likely defend that assumption when asked to review its own work. A different model, with no stake in the original plan, is more likely to push back. The feature is available now via the `/experimental` slash command in GitHub Copilot CLI. Whether it stays useful depends on staying targeted — if it becomes a rubber-stamp ceremony that fires on every trivial change, the signal disappears fast. ## Similar posts on daily.dev - [Rubber Duck in GitHub Copilot CLI](https://daily.dev/posts/rubber-duck-in-github-copilot-cli-h1kpggec0) · The Art of Simplicity · 1 upvotes · 1 comments - [Rubber Duck in GitHub Copilot CLI now supports more models](https://daily.dev/posts/rubber-duck-in-github-copilot-cli-now-supports-more-models-d96bkcovt) · GitHub Changelog · 1 upvotes · 0 comments --- Tags: [#github](https://daily.dev/tags/github), [#claude](https://daily.dev/tags/claude) [View this post on daily.dev](https://daily.dev/posts/github-copilot-cli-gets-a-cross-model-review-agent-called-rubber-duck-bwobwp3qu) ```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":"GitHub Copilot CLI gets a cross-model review agent called Rubber Duck","url":"https://daily.dev/posts/github-copilot-cli-gets-a-cross-model-review-agent-called-rubber-duck-bwobwp3qu","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/github-copilot-cli-gets-a-cross-model-review-agent-called-rubber-duck-bwobwp3qu"},"datePublished":"2026-04-07T23:20:35.875Z","dateModified":"2026-04-16T10:35:15.314Z","description":"GitHub Copilot CLI has added an experimental feature called Rubber Duck that uses a second AI model from a different family to review the primary agent's work...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/1f6f0914fd42ceaf2df22973e341c2b4?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/1f6f0914fd42ceaf2df22973e341c2b4?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Collections","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":"Collections","logo":"https://media.daily.dev/image/upload/s--fk_6ycEi--/f_auto,q_auto/v1780996001/logos/collections?_a=BAMAMiWQ0","url":"https://daily.dev/sources/collections"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/github-copilot-cli-gets-a-cross-model-review-agent-called-rubber-duck-bwobwp3qu","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"github,claude","timeRequired":"PT2M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Collections","item":"https://daily.dev/sources/collections"},{"@type":"ListItem","position":3,"name":"GitHub Copilot CLI gets a cross-model review agent called Rubber Duck"}]} ```