MiniMax introduced M3.1 Flash on Oct. 19, identifying the anonymous Space Bunny model as an early version after it led OpenCode’s daily token usage in September.
MiniMax set prices at $0.15 per million input tokens and $0.60 per million output tokens. Cached input tokens cost $0.015 per million. The company said usage costs for agent tasks are 70% lower than for its predecessor, attributing improved inference efficiency to more aggressive sparsity.
OpenRouter offered the preview under the Space Bunny Alpha name beginning Sept. 23. OpenCode data show it processed 6.8 trillion tokens on Sept. 24, more than any other model on the platform that day. Token usage counts tokens processed, rather than requests. OpenCode’s current table lists the model as M3.1 Flash.
MiniMax said Space Bunny processed more than 200 trillion tokens during its anonymous run. OpenRouter says the free preview period has ended.
MiniMax offers M3.1 Flash through its API, M Plan and MiniMax Code. The company said it would release the model weights later but gave no date.
