Unmasking Space Bunny
After ten days of dominating the OpenRouter platform, the Chinese AI firm MiniMax has confirmed that the anonymous model known as "Space Bunny" was an early build of its new M3.1 Flash model. The model, which officially launched on October 19, processed over 56.7 trillion tokens while operating under its anonymous moniker, according to OpenRouter data.
Competitive Pricing Strategy
MiniMax has positioned M3.1 Flash at $0.15 per million input tokens and $0.60 per million output tokens. This pricing strategy directly matches DeepSeek’s V4.1 Flash off-peak rates, highlighting a defensive move in China’s aggressive market for fast, low-cost AI models. Despite this competitive stance, MiniMax’s financial health remains under pressure; the company reported a net loss of $293 million in the first half of 2026, despite a 283% revenue increase.
Performance and Reliability Concerns
While MiniMax claims the model offers a generational leap in coding and agentic tasks, independent testing paints a more complex picture. Artificial Analysis reported a staggering 93% hallucination rate on its knowledge tests, suggesting the model struggles to identify its own limitations. Furthermore, Plotly co-founder Chris Parmer noted that while the model performs well on benchmarks, it is exceptionally slow in data analysis tasks due to excessive token consumption rather than poor inference speed.
The Road Ahead
MiniMax has promised to release open weights for the model, though as of late October, no files or licenses have appeared on platforms like Hugging Face. The company is also developing a larger Pro model, though no release date has been set. The current state of M3.1 Flash underscores a recurring theme in the industry: high benchmark performance does not necessarily translate into reliable, real-world utility. As MiniMax continues to scale, the industry remains focused on whether the company can bridge the gap between its ambitious marketing claims and the actual performance of its models.
Source: AI Daily News. This story was rewritten and translated for our readers.