Space Bunny Was MiniMax M3.1 Flash

Space Bunny was an anonymous coding model with a 1M-token context window. On October 19, MiniMax confirmed it was an early version of M3.1 Flash, saying it processed more than 200 trillion tokens anonymously. MiniMax claims the new model is nearly 3× faster and cuts agent-task costs by 70%. Official announcement (s'ouvre dans un nouvel onglet)

Both Command Code and OpenCode offered Space Bunny free for a limited time. Command Code has now retired the preview, according to its changelog (s'ouvre dans un nouvel onglet). OpenCode’s model catalog (s'ouvre dans un nouvel onglet) still lists Space Bunny Free at $0, but the offer is temporary, so check availability before relying on it.

M3.1 Flash is now available through MiniMax’s API, M Plan, and MiniMax Code. Published API pricing is $0.15 per million input tokens and $0.60 per million output tokens. That’s not free, but could be affordable for agent workflows if its quality holds up. Pricing and access (s'ouvre dans un nouvel onglet)

Worth watching, not building around a free preview. Test the model on real coding tasks while access lasts, then compare results and cost with your usual model.
You might want to check on this first.

PoorAI·PublicationRéservé aux membresA Cheaper Way to Run AI Coding AgentsA coding tool is trying a different pricing model for developers watching every dollar. The details matter more than the headline.Lire la publication →