Space Bunny Alpha

OpenRouter has launched Space Bunny Alpha, an anonymous large language model that is available for free on its platform. The model went live on September 23 and is offered at no cost for both input and output during its preview. A third-party provider is developing and operating the model while remaining anonymous, with OpenRouter responsible only for routing requests. Space Bunny Alpha supports a 1 million token context window and a maximum output of 524,288 tokens per request. Reported capabilities include fast inference, code generation, adjustable reasoning strength, native multimodal input covering text, images, and video, tool calling, and structured output. The announcement was reported by PANews and distributed through Gate News.

The AI/ML API page confirms Space Bunny Alpha (stealth/space-bunny-alpha) is an anonymous stealth model with fast inference, strong coding, and native multimodal input, always reasoning with adjustable effort over a 1M-token context, free on AI/ML API while cloaked. Pricing is listed at $0.0 per million tokens for both AI/ML API exposes the model through an OpenAI-compatible endpoint at https://api.aimlapi.com/v1/chat/completions using the model id stealth/space-bunny-alpha, so developers can point any OpenAI SDK at it by swapping the base URL. The page specifies a 1,000,000-token input context with up to 524,288 tokens of output, de

On September 23, 2026, an anonymous stealth AI model called Space Bunny Alpha appeared on OpenRouter, Cline, and Command Code, offered free for a one-week testing window under a zero-data-retention policy. Within twelve hours, community researchers matched its tokenizer signatures, Chinese error traces, and prompt defaults to Shanghai-based MiniMax, suggesting the model is an unannounced public test of MiniMax M3. Space Bunny Alpha features a 1,000,000-token context window, a 524,288-token output ceiling, and native text, image, audio, and video processing without separate perception encoders. Developers reported strong procedural coding results, including a single-file Three.js Minecraft clone, but noted formatting inconsistencies and verbosity versus Claude Opus 5.5. The article argues the deployment reflects stealth crowdsourced testing trends and rising Chinese frontier competitiveness in long-output, omnimodal systems.

LM Market Cap's tracker page for stealth Space Bunny Alpha reports a release date of September 23, 2026, with the model listed at $0.00 per 1M input tokens and $0.00 per 1M output tokens. The page reports a composite overall score of 20/100 and a coding category rank of 437 out of 440, based on the site's v3 methodolog Capability signals reported include reasoning, vision, function calling, JSON mode, streaming, and web search support. Input modalities listed are text, image, and video, with text output. The site notes the low composite score may reflect the absence of public benchmarks rather than weakness, and no creator is named b

Siora Labs conducted a direct inspection of OpenRouter's API and confirmed Space Bunny Alpha is served under the model ID stealth/space-bunny-alpha via the /api/v1/chat/completions endpoint. The verbatim description states it is an "anonymous large model with blazing-fast inference, strong coding capabilities and nativ Reasoning is mandatory with effort levels ranging from low through max (defaulting to medium), and supported parameters include reasoning, tools, tool choice, temperature, top_p, max_tokens, response_format, and stream. Pricing is free during the stealth preview, and the model showed 100% uptime over recent intervals.

An independent field-guide website documenting Space Bunny Alpha confirms provider-listed specifications from OpenRouter as of September 24, 2026. The model offers a 1,000,000-token context window with a 524,288-token completion ceiling, and accepts text, image, and video input while producing text output. Both input a The site adds a measured throughput data point of 87 tokens per second at P50 (median) from OpenRouter telemetry, a metric not present in other sources. The site is self-described as an independent community reference and explicitly notes the model operator has not been publicly identified. It links back to OpenRouter'