Accessibility Adjustments

Use these optional tools to adjust reading and display preferences. These tools cannot resolve every accessibility barrier. Please contact the website owner if you need assistance.

  • Text adjustments
  • Content scaling 100%
  • Font size 100%
  • Line height 100%
  • Letter spacing 100%
  • Colour adjustments
  • Orientation adjustments

Space Bunny remains listed on OpenCode after its OpenRouter preview ends

Listen to this article

Space Bunny Alpha’s OpenRouter preview ended on October 5, 2026. That announcement said its identity would be revealed soon. It did not say that every other service would stop offering access.

On October 6, OpenCode announced its own extension. It was sponsoring a few more free days while the model was being tuned before a reveal.

The remaining question is which route a particular application uses, what it charges and whether it can still complete a request. This availability check was made on October 13, 2026.

What the current catalogs show

OpenRouter’s provider endpoint lists no available providers for Space Bunny Alpha. The model is also missing from its active catalog. Its descriptive webpage remains online, but that page alone does not establish usable access.

OpenCode’s public Console model catalog still includes Space Bunny Free. Its model documentation describes Space Bunny Free as available without token charges for a limited period.

The Go documentation separately lists Space Bunny at $0.15 per million input tokens, $0.60 for output and $0.03 for cached input. The published monthly model allowance is $30 on Go and $120 on Go Plus. Those allowances describe included model usage, not the subscription price.

A familiar name in a selector is not enough to tell you which billing arrangement you selected. Compare the actual destination and its current terms before deciding how much work to send.

A catalog check is not an inference test

ByteForward checked the public listings and announcements for this article. We did not submit an authenticated generation request. Catalog entries alone do not guarantee that every account can use a model today.

The October 6 extension also should not be read as a permanent commitment. Its wording covered a few more days and gave no replacement closing date.

For a developer moving an existing project, there are two separate questions. Can the destination currently serve the requested model, and does the account have access under acceptable terms? Answering only the first can leave an integration that looks correct in configuration but fails when work begins.

What the listings say about identity

OpenRouter’s historical model description still attributes Space Bunny Alpha to an anonymous third party. OpenRouter describes itself as the router rather than the developer. The published preview advertised a context window of one million tokens.

A shared codename across services does not prove that both routes use an identical checkpoint, configuration or serving setup. Nor can a model’s answer about its own identity authenticate its maker. A definitive attribution needs a statement from the developer or a platform that identifies the model it serves.

That uncertainty matters when comparing saved results. If an earlier coding test used one gateway and a new trial uses another, record the change. Otherwise, a difference in output may be incorrectly attributed to a newly revealed model or a quality improvement that nobody has established.

Check the route before moving a project

Start with the application’s current provider and model selection. Save those settings before changing anything. Keep the provider name, exact model identifier and request destination together so that a working configuration can be restored.

Next, inspect the destination’s current model list and price. Do not copy an old free label from a tutorial into a spending assumption. If the destination offers both free and paid options, choose deliberately and set a limit appropriate to a small trial.

Use a short prompt with public or synthetic material for the first request. Confirm that an answer arrives and check the usage record before submitting a long conversation or repository. Preserve the actual error if it fails. That makes a support request more useful than repeatedly changing unrelated settings.

Only then compare a real task. Keep the prompt and expected result the same, and record the route and date alongside the outcome. For coding work, retain the original failing test and run the existing test suite after reviewing any proposed patch.

Choose a fallback before the next interruption

For a workflow that must run reliably, keep the model selection configurable and test a replacement before the preferred option becomes unavailable. Judge the fallback on the tasks that matter to the project, including its failure behavior and the cost of repeated attempts.

ByteForward’s guide to OpenRouter router benchmarks provides related context for comparing quality, time and cost. A benchmark can help narrow a shortlist, but it cannot confirm that a particular account has access to a route today.

For a project that depends on this model, verify the chosen route, document its current terms and keep a tested fallback ready.

Illustrative coding photograph by Daniil Komov via Pexels, used under the Pexels License. The photograph does not show OpenCode or a Space Bunny test.

Jordan Reid
Jordan Reid

Jordan Reid is focused on AI tools, agents, developer products, and the way technology changes everyday work. Jordan approaches a launch from the user’s side of the screen. What can it actually help someone finish? The voice is practical, conversational, and skeptical of products that turn a simple job into five new settings. Coverage follows coding assistants, creative software, browser agents, and the workflows around them, with attention to pricing, permissions, setup, and the human work that remains.

Leave a Reply

Your email address will not be published. Required fields are marked *

Gravatar profile