SpaceBunny

SPACE BUNNY / FEATURE GUIDE

Space Bunny Alpha video understanding archive

Alpha once accepted video and returned text. Native video is currently unavailable after its retirement. This page preserves our original MP4 and WebM observations and explains how to use selected image frames with the current DeepSeek chat.

Space Bunny Alpha video understanding features

Open the free workspace ↗

How to use it

  1. Choose up to three representative frames from your video on your device. Native video upload is unavailable in this current integration.
  2. Open the chat and attach up to three PNG, JPG or WebP screenshots, with a combined upload size of no more than 2.5 MB.
  3. Label each frame with its timestamp in your question. Describe any action between frames that the screenshots cannot show.
  4. Ask DeepSeek V4.1 Flash to explain the visible sequence and identify uncertainty. Cross-check its answer against the original clip.

Try this prompt

Summarize the visible steps in this demonstration. List the actions in order, explain where the process changes, and say which details cannot be confirmed from the video.
Use this prompt in chat →

Archived Alpha MP4 video test

In a local test, uploading this 7 KB MP4 returned the order red, blue, green. A WebM version failed, during the original Alpha preview. Native video is currently unavailable with DeepSeek V4.1 Flash. The clip demonstrates sequence understanding; the model did not generate it.

When it helps

Try a short product demonstration, a scene explanation, or a process walkthrough. Keep the clip focused on the part you need help with. Avoid uploading footage containing private or sensitive information. Attachments are stored with your chat until you delete that conversation; provider retention is separate.

Limits and checks

Native MP4 and video URL analysis is unavailable after the Alpha model's withdrawal. The screenshot workflow can analyze visible frames but cannot watch the full clip, hear audio, transcribe speech or reliably infer off-frame actions. Do not upload private footage. The old MP4 test below is an archived Alpha observation, not a test of the current model's video support.

Frequently asked questions

Can I upload a video to this site now?

No. Space Bunny Alpha retired October 5, 2026. Current DeepSeek V4.1 Flash chat supports text and images, not native video or video URLs. Use selected screenshots instead.

Why did MP4 work better than WebM in the old test?

The original Alpha test read red, blue and green from a 7 KB MP4, while the same sequence in WebM failed. This was a format-specific observation during the preview, not proof that all MP4 clips work or a verified explanation of the provider's decoder.

What if my attachments exceed 2.5 MB?

For the current screenshot workflow, resize or compress the images and keep the combined upload under 2.5 MB, with at most three images. Video compression does not restore native video support in the current model.

Can this chat transcribe speech or hear audio?

No verified audio or transcription workflow is offered here. Selected screenshots contain no audio; paste your own transcript as text if you need help discussing it.

What should I do when a video request fails?

Do not keep retrying native video: the current model does not support it. Start a text or image conversation and attach key frames. For an image error, check type and size. For 429, wait; for 402, contact support so the operator can check upstream credits.

Can screenshots establish the order of every action?

No. Label each frame with its timestamp and explain missing intervals. The model can discuss visible frames but cannot reliably reconstruct unseen motion, exact timing, audio or events outside the frame.

Does the video example generate a clip?

No. The example is a fixed local test asset used during Alpha's preview. The model described it in text; neither the retired preview nor this app's current workflow is video generation.

Space Bunny Alpha: what it is, how to get in and app differences →

Keep exploring

Capability reference: OpenRouter ↗