Xiaomi challenges ai giants with surprisingly powerful, cheap llm

Xiaomi, best known for its aggressively priced smartphones, is quietly launching a serious bid to disrupt the artificial intelligence landscape. Its new large language model, MiMo-V2-Pro, isn't just another entrant; it’s undercutting established players like Anthropic and OpenAI on both performance and cost, potentially reshaping the economics of AI development and access.

The benchmark blowout: mimo-v2-pro exceeds expectations

Initially released in March, MiMo-V2-Pro generated modest buzz. However, recent benchmark tests reveal a dramatic shift. Xiaomi claims the model consistently outperforms Claude Sonnet 4.6, Gemini 3 Pro, and even rivals GPT-5.2 in key areas. The sheer speed of this improvement—moving from a promising but underwhelming debut to a top-tier contender in a matter of months—is noteworthy in itself.

What's driving this performance? MiMo-V2-Pro boasts 42 billion actively utilized parameters within a model exceeding a trillion total, capable of processing up to one million tokens – a level competitive with leading AI models. But here's the kicker: Xiaomi is offering this power at a fraction of the price.

Cost-effective ai: a game changer?

The pricing structure is particularly compelling. Charging between $3 and $6 per million tokens for output, MiMo-V2-Pro significantly undercuts Anthropic's Claude Sonnet 4.6 and Opus 4.6, which command prices ranging from $3 to $5 for input and a steep $15 to $25 for output. The cost for initial input is similarly aggressive, with a $1 per million token option offering 256,000 tokens of context, while a $2 per million token plan offers up to 1,000,000 tokens. This represents a fundamental shift in the AI pricing model, potentially democratizing access for developers and smaller companies.

The implications extend beyond mere affordability. Xiaomi’s strategy of integrating MiMo-V2-Pro with OpenClaw, an open-source AI automation framework, is a direct challenge to Anthropic’s recent decision to restrict Claude’s integration with third-party tools due to token consumption concerns. Xiaomi is essentially betting that open collaboration and affordable access will prove more valuable than proprietary control.

Beyond the hype: a mobile giant's unexpected pivot

While Xiaomi’s foray into AI might seem like a sudden shift, it aligns with the company’s broader strategy of aggressive expansion and vertical integration. Selling a smartphone for as little as €75 while simultaneously offering cutting-edge AI capabilities demonstrates a willingness to disrupt established norms. MiMo-V2-Pro, initially offered for free during its introductory period and now available via API, represents a calculated risk with the potential for substantial returns. The fact that they can offer this power at such a low cost, while delivering comparable performance in programming and general tasks, speaks volumes about their engineering prowess.

The move signals that Xiaomi is more than just a smartphone manufacturer; it's an emerging force in the AI arena, and its pricing strategy could force the hand of industry leaders.