STATUS UPDATED / OCTOBER 6, 2026
How to Use Space Bunny Alpha
The upstream Space Bunny Alpha preview ended October 5, 2026. The steps below still describe this site's chat interface, which now connects through OpenRouter to DeepSeek V4.1 Flash. Text, coding and image understanding remain available with 50 successful free replies per UTC day; native video is unavailable.
The Alpha preview has ended. Chat here uses DeepSeek V4.1 Flash; existing free sign-in and daily allowance apply.
Step 1: Open the chat and sign in
Select “Start chat” on this page. If you are signed out, the site will ask you to sign in before it sends a request. Continue with Google, or create an email account and verify the message sent to your inbox. If the email signup allowance is full for the day, Google sign-in remains the alternative. You can read public guides without an account, but sending chat messages is authenticated.
A heading that promises “no login” would be inaccurate for this site. Sign-in keeps conversations attached to your account and helps enforce a fair daily allowance across the shared preview. No card, top-up, or paid subscription is part of this first release.
Step 2: Start with a focused text question
Write a concrete task with relevant code, the exact error and expected behavior. Replies now come from DeepSeek V4.1 Flash through OpenRouter, not the retired Space Bunny Alpha or a model trained by this website. Verify important facts and code before using them.
This preview allows up to 20,000 characters per message, keeps 20 recent context messages, and uses an 8,192-token output budget including reasoning. Those are workspace limits, not claims about the model's full context window. A multi-turn conversation can carry a follow-up question, but earlier details may leave the retained context as the conversation grows.
Step 3: attach images when useful
Attach up to three images within a combined 2.5 MB budget. Name the visible detail, object or printed text you want to inspect. Current chat does not accept MP4 or video URLs and does not generate images, video or speech.
Historical Alpha tests identified shapes and ORBIT-731 and described red, blue and green in a small MP4; WebM failed. These observations were recorded while the preview was live and do not establish video support in the current model. Use representative screenshots instead.
Step 4: Choose controls only when they help
The composer offers adjustable reasoning effort, a JSON output option, and bounded tools for arithmetic, unit conversion, and current time. Choose a higher reasoning effort for difficult reasoning tasks, but expect it to spend more of the output budget. JSON output helps with structured notes; do not assume a strict schema will always be enforced. The site's tools do not provide open web search or arbitrary code execution.
Space Bunny Alpha was withdrawn upstream on October 5, 2026. OpenRouter now lists no available endpoints, and a direct API check on October 6 returned 404. This site continues with DeepSeek V4.1 Flash for text, coding and image understanding. It does not currently process video input. Alpha's maker has not been confirmed; no official successor or reveal date has been established.
What to do when something fails
Start a fresh login flow rather than reusing a callback. Wait for the UTC reset if the daily allowance is exhausted. For images, check the format and combined size. Video is currently unsupported; use screenshots. A 402 requires the operator to check upstream credits; use bounded backoff for 429 instead of rapid retries.
Alpha retired on October 5 and this site now explicitly uses DeepSeek V4.1 Flash. Changing models does not preserve the original model capabilities. Avoid private material; report the error text to support without sending keys or passwords.
Frequently asked questions
Can I chat without logging in?
No. Public guide pages are open, but this site's chat requires Google sign-in or a verified email account. No payment card is needed.
How many free replies can I send?
The default is 50 successful replies per account per UTC day. The allowance resets at 00:00 UTC and can change if the preview configuration changes.
Why did my video upload fail?
The original model supported text, image and video input and returned text. It has been withdrawn. This site's DeepSeek V4.1 Flash integration accepts text and images, currently does not accept video, and does not generate images, video or speech.
Can the model create images or spoken audio here?
The original model supported text, image and video input and returned text. It has been withdrawn. This site's DeepSeek V4.1 Flash integration accepts text and images, currently does not accept video, and does not generate images, video or speech.
What if Google login sends me back to the homepage?
Start a fresh sign-in from the login page. If the problem persists, use verified email login and contact support with the error text, without sharing a password or token.
Why does JSON not match my exact schema?
The model listing supports a JSON response format but does not promise JSON Schema enforcement. Validate the result in your own workflow.
Sources and updates
Upstream availability can change without notice; this site's allowance and upload limits come from its current implementation.
This site's chat runs on DeepSeek V4.1 Flash. Free sign-in and daily allowance apply.