AI Desk · AI FingerprintingNew arrival
Why high: an exact fingerprint shared with 8 models.
Independent YFarmX measurements indicate that Space Bunny Alpha is consistent with the MiniMax family at high confidence. These measurements establish lineage evidence only: a shared tokenizer or serving contract shows shared ancestry, and does not independently establish the precise underlying release.
Listed anonymously at 14:48 UTC on 23 September 2026 and measured the same day. Its token counts are MiniMax's on all 50 strings: every one of the eight MiniMax models in the dataset, from MiniMax-01 to M3, returns the same 50 counts, against 29 of 50 for the next closest family (Moonshot's Kimi K2) and a median of 11 across 287 measured models. Two passes agreed on every string and the fixed overhead held at 143 tokens before and after. MiniMax models share one vocabulary, so the counts place the family and cannot separate its versions. The listing has MiniMax M3's shape: a 1M-token context with text, image and video input, the only MiniMax model with that combination. Our read is a MiniMax model newer than M3. Asked who built it, the endpoint answers OpenAI and GPT-5, a claim the counts exclude: no OpenAI model uses this vocabulary. Raw measurements and the working are in docs/research/space-bunny-alpha/.
Space Bunny Alpha on OpenRouter (stealth listing, 23 September 2026)MiniMax M3 on OpenRouter
| We checked | What we found | Where it came from |
|---|---|---|
| How it counts tokens | Counts every one of the 50 test strings exactly like 8 other models | tk_9f84b5fb |
| How its API is set up | A combination of settings no other listing in the catalogue uses | api_dd9fb93d |
| How much it can read and write | Reads up to 1,000,000 tokens at once, writes up to 524,288 | its own listing |
| Whether it thinks before answering | Always reasons, and cannot be turned off, at max, xhigh, high, medium, low, max by default | its own listing |
| Who served our requests | Stealth | we measured it, 23 September 2026 |
Click a row to open the full comparison.
We sent Space Bunny Alpha fifty short pieces of text and recorded what each one cost it in tokens. The bars below are those costs. Two models built on the same tokenizer produce the same bars; a model built on a different one produces a different set, which is what makes this a fingerprint.
Overhead subtracted: 143 prompt tokens.
Space Bunny Alpha is an anonymous large model with blazing-fast inference, strong coding capabilities and native multimodal input support. It delivers adjustable reasoning effort, and a 1M-token context window. Space...
Supported parameters
include_reasoningmax_tokensreasoningreasoning_effortresponse_formattemperaturetool_choicetoolstop_p
defaults: {}
