Xiaomi challenges ai titans with surprisingly agile llm
Xiaomi, best known for its aggressively priced smartphones, is making a surprisingly forceful entry into the artificial intelligence arena. While the company has dabbled in AI features before, the release of its MiMo-V2-Pro language model signals a far more ambitious play for dominance in a field currently dominated by Anthropic's Claude and OpenAI's ChatGPT.
Early benchmarks show unexpected performance gains
Initially launched in March, MiMo-V2-Pro garnered some attention but didn't immediately threaten the established leaders. However, recent benchmark analyses reveal a dramatic shift. Xiaomi claims its model is now highly competitive, even surpassing Claude Sonnet 4.6, Gemini 3 Pro, and GPT-5.2 in several key areas. The company insists preliminary versions of MiMo-V2-Pro already outperformed Sonnet 4.6 in a majority of scenarios, a claim substantiated by their internal testing.
What’s truly remarkable is the sheer efficiency of the model. MiMo-V2-Pro boasts over a billion parameters, with 42 billion actively engaged, capable of handling up to one million tokens—a level of complexity found in many of the most advanced AI models today. But here's the twist: Xiaomi is achieving this performance at a fraction of the cost.
The cost advantage: a disruptive factor
The economics are compelling. MiMo-V2-Pro is significantly cheaper than competitors like Claude Sonnet 4.6 and Opus 4.6, delivering comparable performance in both programming and general tasks. During its initial rollout, MiMo-V2-Pro was even offered entirely free of charge. Now, pricing starts at $2 per million tokens for input with a 1 million token context window, or $1 per million tokens with a 256,000 token context limit, culminating in an output cost of $3-$6 per million tokens. Anthropic’s models, by comparison, range from $3 to $5 per million input tokens and a staggering $15 to $25 per million output tokens.
Consider this: Xiaomi sells a smartphone for roughly €75, and has integrated MiMo-V2-Pro to power OpenClaw, a desktop Linux distribution. This demonstrates a strategy of democratizing access to powerful AI, a stark contrast to Anthropic's recent decision to restrict Claude’s integration with third-party tools like OpenClaw, citing excessive token consumption.
A bold strategy: open source and accessibility
While Anthropic tightens its control, Xiaomi is effectively opening the floodgates. By supporting OpenClaw, the company is betting on a future where powerful AI models are readily accessible and integrated into a wider range of applications. The move isn't just about competing with Claude; it's about shaping the ecosystem and positioning Xiaomi as a champion of open-source AI development.
The implications are clear: Xiaomi isn't just building smartphones anymore. It’s building a foundation for a new generation of AI-powered tools, grounded in a philosophy of affordability and accessibility. The initial skepticism surrounding MiMo-V2-Pro has rapidly evaporated, replaced by a growing recognition of Xiaomi's potential to disrupt the AI landscape—not with brute force, but with clever engineering and a shrewd understanding of market dynamics.