Xiaomi's MiMo models are making the AI pricing conversation uncomfortable

*Xiaomi's MiMo Models Disrupt AI Pricing Landscape*

Xiaomi's recent release of the MiMo-V2-Flash and MiMo-V2-Pro models has sparked a conversation around AI pricing that is becoming increasingly uncomfortable for Western AI companies. The open-source nature of these models, combined with their competitive pricing and performance, is challenging the traditional business model of high-end AI solutions.

Open-Source and Affordable

The MiMo-V2-Flash model, in particular, has gained attention for its impressive performance on the SWE-Bench, scoring 73.4% and ranking #1 among open-source models. What's more, it costs a mere $0.10 per million input tokens, a price point that is comparable to Claude Sonnet at 3.5% of the cost. This affordability factor is a significant departure from the high-end AI solutions that typically come with a hefty price tag.

Competition from the East

The MiMo-V2-Pro model, on the other hand, has proven to be a formidable competitor in the agent benchmarks, ranking #3 globally behind Claude Opus 4.6. With a 1M token context window, it offers similar performance to Opus, but at a significantly lower price point of $1/$3 per million tokens. This is a stark contrast to Opus, which charges $5/$25 for similar performance.

The Deep Connection

Interestingly, the lead researcher behind the MiMo-V2-Pro model has a background with DeepSeek, a company known for its expertise in AI development. The fact that the Pro model was able to spend a week on OpenRouter anonymously without being detected suggests that it may have been developed using resources and expertise from DeepSeek. This raises questions about the true nature of the MiMo-V2-Pro model and its potential implications for the AI industry.

Time for a Response?

The success of Xiaomi's MiMo models has left Western AI companies scrambling to respond to the new pricing dynamics. While some may argue that the high-end AI solutions offer superior reliability, safety, and enterprise support, the 10x premium that comes with it may not be justified in the face of open-source, affordable alternatives. At what point will Western AI companies be forced to reconsider their pricing strategies and respond to the growing competition from the East? Only time will tell, but one thing is certain: the AI pricing conversation has become increasingly uncomfortable for those who have traditionally dominated the market.