Space Bunny Alpha in Claude Code
Claude Code can drive any OpenAI-compatible model. Set the base URL to our hosted endpoint, pick space-bunny-alpha, and you get a fast, 1M-context model — free during the OpenRouter preview — reading your whole repo in one session.
Every example needs a free sb_live_ key. Get one free →
Space Bunny Alpha is an OpenAI Chat Completions-compatible chat model with a 1,000,000-token context window, strong coding, and very fast inference — a good match for an agent that has to hold a lot of code in view. Because Claude Code accepts any OpenAI-compatible endpoint, you don't need a special plugin: you point it at our hosted API, name the model space-bunny-alpha, and Claude Code talks to it like any other backend.
Get a key
Our API is OpenAI-compatible and metered per key. Grab a free sb_live_ key at /get-jev, then export it so nothing sensitive lands in your config or source.
export SPACE_BUNNY_API_KEY=sb_live_...Point Claude Code at the endpoint
Claude Code reads an OpenAI-compatible base URL and API key from its environment / settings. Set the base URL to our gateway and the model to space-bunny-alpha:
# OpenAI-compatible backend
OPENAI_BASE_URL=https://spacebunnymodel.com/api/v1
OPENAI_API_KEY=sb_live_...
# then select the model:
# space-bunny-alphaExact variable names vary by Claude Code version and setup — the constants are the base URL (https://spacebunnymodel.com/api/v1), the sb_live_ key, and the model name space-bunny-alpha. If your build wires custom models differently, the how-to-use page has the current, copy-paste config. Requests land on POST https://spacebunnymodel.com/api/v1/chat/completions — the standard chat-completions surface.
Why it's good in Claude Code
- 1M-token context: point Claude Code at the whole repo instead of a handful of files — Space Bunny can hold it, which is where a coding agent gets most of its leverage.
- It's a reasoning model with adjustable reasoning: give it room (a higher max output) on hard refactors and let it think before it edits.
- Very fast and free during the preview, so high-volume agent loops don't run up a bill today.
- Native multimodal input (text, image, audio, video) upstream, if your Claude Code workflow attaches it.
No custom-model support in your build?
If your Claude Code version doesn't expose a base-URL override, put a small OpenAI-compatible proxy in front and route it to https://spacebunnymodel.com/api/v1 with your sb_live_ key. Either way the model name stays space-bunny-alpha and the wire format is plain chat completions.
FAQ
How do I use Space Bunny Alpha in Claude Code?
Set Claude Code's OpenAI-compatible base URL to https://spacebunnymodel.com/api/v1, use a free sb_live_ key, and select the model space-bunny-alpha. Requests go to the standard /api/v1/chat/completions endpoint — see /how-to-use for the exact current config.
Do I need a paid key?
No — grab a free sb_live_ key at /get-jev. There's a free tier plus credit packs if you need higher limits, and the underlying model is free during the OpenRouter preview.
Why Space Bunny Alpha for coding agents?
The 1,000,000-token context lets Claude Code load a whole codebase at once, it's a fast reasoning model with strong coding, and it's free during preview — so iterative agent loops stay cheap and fast.
What if my Claude Code build has no base-URL override?
Run a small OpenAI-compatible proxy pointed at https://spacebunnymodel.com/api/v1 with your sb_live_ key, and have Claude Code talk to the proxy. The model name stays space-bunny-alpha.
See also: Get a free API key · How to call the API · opencode integration · Cursor integration
Start using Space Bunny Alpha
Try it free in the browser, then grab a key and drop it into your setup.