Space Bunny Free

Space Bunny Free appears in the OpenCode catalog as an anonymous, limited-time free model, distributed under the identifier space-bunny-free with no disclosed underlying family, parameter count, or training lineage. According to a third-party promotional write-up about the listing, the offering is positioned for hands-on experimentation rather than as a permanent production tier, and it is made accessible to both Zen account holders and Go subscribers at no charge to their Go quota. The framing of the model as anonymous means creators cannot be attributed from the available evidence, so the practical takeaway is that users encounter it as a short-window, no-cost option inside the OpenCode environment rather than a documented release from a known lab.

From a capabilities standpoint, the listing advertises multimodal input alongside an approximately one-million-token context window and a zero-data-retention promise from the provider, which together suit long-context, mixed-media tasks such as extended document review, code-and-image pairing, or video-aware reasoning during the trial period. Because no benchmark scores, architecture notes, or post-training methodology have been published, the model is best approached as a general-purpose workhorse for opportunistic evaluation: it is convenient for developers who already use OpenCode and want to stress-test very large contexts or multimodal flows without burning through quota, but it should not be relied on for committed pipelines until independent technical disclosure emerges.