--- title: Space Bunny Alpha: what the new stealth model on... description: A new stealth AI model called Space Bunny Alpha quietly appeared on OpenRouter and OpenCode, free in OpenCode for a limited week. It offers a 1 million token... canonical: https://daily.dev/posts/space-bunny-alpha-what-the-new-stealth-model-on-openrouter-can-actually-do-rjj4m5ig9 twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Space Bunny Alpha: what the new stealth model on OpenRouter can actually do | daily.dev og:description: A new stealth AI model called Space Bunny Alpha quietly appeared on OpenRouter and OpenCode, free in OpenCode for a limited week. It offers a 1 million token... og:url: https://daily.dev/posts/space-bunny-alpha-what-the-new-stealth-model-on-openrouter-can-actually-do-rjj4m5ig9 og:image: https://api.daily.dev/og/posts/rjJ4M5iG9.png og:image:alt: Space Bunny Alpha: what the new stealth model on OpenRouter can actually do 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. # Space Bunny Alpha: what the new stealth model on OpenRouter can actually do **[Collections](https://daily.dev/sources/collections)** · 3 min read · 1 upvotes · 0 comments ## Summary A new stealth AI model called Space Bunny Alpha quietly appeared on OpenRouter and OpenCode, free in OpenCode for a limited week. It offers a 1 million token context window and native multimodal input handling text and images together, with reportedly very fast inference. Testers found it excels at image-to-code tasks - turning wireframes into landing pages, photos into 3D models, and hand-drawn sketches into playable Three.js games - plus a notable self-verification behavior where it screenshots and fixes its own output before finishing. It struggles with physics simulation, losing to GLM-5.3 on a Newton's cradle test and failing tornado and water drop scenes alongside it. Nobody has confirmed who built the model or what's underneath. ## Content A new model called Space Bunny Alpha quietly launched this week on OpenRouter and OpenCode. It's currently in stealth - nobody's confirmed who built it or what's underneath - but it's free in OpenCode for the next week, so people have been putting it through its paces. The specs are straightforward: 1 million token context window, native multimodal input (text and images in the same flow), and reportedly fast inference - described by one tester as "flash-level, really unexpected." ## What it's good at Omar Sanseviero ran a focused set of visual tests, starting from single images or rough sketches. A few results stood out: - From a wireframe with seven handwritten notes, it built a complete landing page that followed every note. - From a single photo of a moka pot, it produced a 3D model that matched the eight-sided body, brass valve, and camping stove underneath. - It built a labeled 3D anatomy of a mirrorless camera with 67 parts, an 11-element lens, a 9-blade iris, and an exploded view. - Given a hand-drawn game level sketch with ten written rules, it built a playable Three.js game, then opened Chrome, playtested its own jumps, and tuned the physics until every lava stone was reachable. - For a 60-second animated proof of why a circle's area is πr², it generated slices that rearrange into a rectangle, a live slider, captions, and narration. Not perfect, but functional. The self-verification behavior is worth noting separately. When it has browser access, it opens the page, looks at screenshots, and fixes what it sees before calling the task done. That kind of feedback loop matters for longer agentic tasks. Another tester dropped in an image, asked what it saw, and got an immediate accurate response - text and image handled in the same prompt flow without friction. ## Where it struggles Physics simulation is a weak spot. In a five-scene test run by AI/ML API comparing Space Bunny Alpha against GLM-5.3, GLM-5.3 won on Newton's cradle. Both models struggled with tornado and water drop simulations. The underlying issue is that visual quality can't compensate for wrong collision timing or momentum transfer - if the physics logic is off, adding more environmental detail doesn't fix it. ## Bottom line For image-to-code work, interactive prototypes, and creative 3D coding, it's worth trying while it's free. The speed and the self-checking behavior are the two things that seem to genuinely differentiate it from what's currently available. Physics simulation is not where you'd reach for it. ## Questions this post answers ### What is Space Bunny Alpha and what can it do on OpenRouter and OpenCode? Space Bunny Alpha is a stealth AI model that appeared on OpenRouter and OpenCode, free to use in OpenCode for a limited week. It has a 1 million token context window, native multimodal text-and-image input, and fast inference described as flash-level. It excels at turning wireframes, sketches, and photos into working code, landing pages, 3D models, and playable Three.js games, and it self-verifies by screenshotting its own output before finishing. _Developers testing new stealth models can track capability reports like this on daily.dev._ ### Is Space Bunny Alpha good at physics simulation compared to other AI models? No, physics simulation is a weak spot for Space Bunny Alpha. In a five-scene comparison against GLM-5.3, GLM-5.3 won the Newton's cradle test, and both models struggled with tornado and water drop simulations because wrong collision timing or momentum transfer can't be fixed by adding more visual detail. _Anyone choosing an AI model for simulation-heavy coding tasks can follow comparisons like this on daily.dev._ ## Similar posts on daily.dev - [A mystery AI model has developers buzzing](https://daily.dev/posts/a-mystery-ai-model-has-developers-buzzing-8adj9fnlg) · TechCentral · 1 upvotes · 0 comments --- Tags: [#llm](https://daily.dev/tags/llm), [#ai-agents](https://daily.dev/tags/ai-agents), [#multimodal](https://daily.dev/tags/multimodal), [#opencode](https://daily.dev/tags/opencode), [#openrouter](https://daily.dev/tags/openrouter) [View this post on daily.dev](https://daily.dev/posts/space-bunny-alpha-what-the-new-stealth-model-on-openrouter-can-actually-do-rjj4m5ig9) ```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":"Space Bunny Alpha: what the new stealth model on OpenRouter can actually do","url":"https://daily.dev/posts/space-bunny-alpha-what-the-new-stealth-model-on-openrouter-can-actually-do-rjj4m5ig9","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/space-bunny-alpha-what-the-new-stealth-model-on-openrouter-can-actually-do-rjj4m5ig9"},"datePublished":"2026-09-28T16:24:47.382Z","dateModified":"2026-09-28T16:25:30.866Z","description":"A new stealth AI model called Space Bunny Alpha quietly appeared on OpenRouter and OpenCode, free in OpenCode for a limited week. It offers a 1 million token...","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/space-bunny-alpha-what-the-new-stealth-model-on-openrouter-can-actually-do-rjj4m5ig9","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"llm,ai-agents,multimodal,opencode,openrouter","timeRequired":"PT3M"} {"@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":"Space Bunny Alpha: what the new stealth model on OpenRouter can actually do"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/space-bunny-alpha-what-the-new-stealth-model-on-openrouter-can-actually-do-rjj4m5ig9#faq","mainEntity":[{"@type":"Question","name":"What is Space Bunny Alpha and what can it do on OpenRouter and OpenCode?","acceptedAnswer":{"@type":"Answer","text":"Space Bunny Alpha is a stealth AI model that appeared on OpenRouter and OpenCode, free to use in OpenCode for a limited week. It has a 1 million token context window, native multimodal text-and-image input, and fast inference described as flash-level. It excels at turning wireframes, sketches, and photos into working code, landing pages, 3D models, and playable Three.js games, and it self-verifies by screenshotting its own output before finishing. Developers testing new stealth models can track capability reports like this on daily.dev."}},{"@type":"Question","name":"Is Space Bunny Alpha good at physics simulation compared to other AI models?","acceptedAnswer":{"@type":"Answer","text":"No, physics simulation is a weak spot for Space Bunny Alpha. In a five-scene comparison against GLM-5.3, GLM-5.3 won the Newton's cradle test, and both models struggled with tornado and water drop simulations because wrong collision timing or momentum transfer can't be fixed by adding more visual detail. Anyone choosing an AI model for simulation-heavy coding tasks can follow comparisons like this on daily.dev."}}]} ```