A stealth model gaining traction across coding agents — a 1,000,000-token context window, multimodal input, and strong coding performance, served fast. Try it free in your browser, or drop it into your app via our API.
A stealth preview model with frontier-scale context, served fast — and free while it lasts.
Drop a whole repo, a long transcript, or a book into a single call. 1M tokens in, up to 524k out — no chunking gymnastics.
Space Bunny Alpha is free right now. Try it in the browser with zero signup, or point your code at our hosted endpoint.
Adjustable reasoning effort, native text-vision-audio-video input, strong coding, and blazing-fast inference.
Most models cap out long before a real codebase or document fits. Space Bunny Alpha takes 1,000,000 tokens in a single call — and returns up to 524,288.
Free in the browser, a drop-in hosted API, straight from OpenRouter, or as a 1M-context tool inside your agent — your call.
The fastest way to code with Space Bunny for free. Point opencode at our OpenAI-compatible endpoint and go.
opencode setup →One OpenAI-compatible key, no OpenRouter account, no rate-limit surprises. Drop it into any SDK, CLI or agent.
Get an API key →Prefer the source? Call stealth/space-bunny-alpha directly. We show you exactly how, plus the caveats.
How to call it →Give Claude Code, Cursor or opencode a 1M-context tool via MCP — offload whole-repo analysis to Space Bunny.
Install the tool →Space Bunny Alpha is an anonymous ('stealth') large language model available on OpenRouter since September 2026. It has a 1M-token context window, native multimodal input, adjustable reasoning, and strong coding — served fast and, during the preview, for free.
The provider is anonymous during the preview. Community reverse-engineering of its tokenizer and behavior points to MiniMax M3, but the identity is unconfirmed. OpenRouter only routes requests to it.
Yes — it's free during the OpenRouter preview. Our browser playground is free with no signup. Our hosted API has a free tier plus paid plans for higher limits and reliability when the model graduates to paid.
Point opencode's OpenAI-compatible provider at our endpoint (or OpenRouter) and select space-bunny-alpha. Full step-by-step on our opencode integration page.
1,000,000 input tokens and up to 524,288 output tokens — one of the largest available, ideal for whole-codebase and long-document tasks.
Free, no signup, streamed from the real API. Then grab a key to use it anywhere.
Open the playground ▶