Xiaomi challenges ai giants with surprisingly affordable llm
Xiaomi, often recognized for its competitive smartphones, is making a far bolder play: directly challenging established leaders in the artificial intelligence arena. The company's newly unveiled MiMo-V2-Pro language model isn't just another entrant; early benchmarks suggest it’s delivering performance rivaling significantly more expensive offerings from Anthropic and OpenAI, all while undercutting their pricing.
The unexpected contender
The LLM, launched in March, initially garnered modest attention. But recent testing reveals a startling shift. Xiaomi claims MiMo-V2-Pro now consistently outperforms Claude Sonnet 4.6 and approaches the capabilities of Gemini 3 Pro and even GPT-5.2 in key areas. The model, designed to power AI agents, boasts a staggering 42 billion always-active parameters, leveraging up to a million tokens – a scale comparable to many industry-leading models.
What truly sets MiMo-V2-Pro apart is its cost-effectiveness. While Anthropic's Claude Sonnet 4.6 and Opus 4.6 command prices ranging from $3 to $5 per million input tokens and a hefty $15 to $25 per million output tokens, Xiaomi offers MiMo-V2-Pro for between $3 and $6 per million tokens of output. The initial release even provided the model entirely free of charge, though it now operates on a token-based API pricing structure.
Consider this: Xiaomi sells smartphones for as little as €75, often with 128GB of storage, and leverages MiMo-V2-Pro to propel OpenClaw, a move that underscores the company's ambition to democratize access to powerful AI tools. Anthropic recently restricted Claude's integration with third-party utilities like OpenClaw, citing excessive token consumption. Xiaomi, conversely, is actively embracing such integration, revealing a clear strategic divergence.
Beyond price: a potential ecosystem play
The implications extend beyond mere cost savings. Xiaomi’s move positions it uniquely to foster a vibrant ecosystem around OpenClaw and other AI applications. By offering a competitively priced and powerful LLM, they're lowering the barrier to entry for developers and users alike. The decision to support OpenClaw, while frustrating Anthropic’s existing user base, signals Xiaomi’s willingness to disrupt the status quo. The user cost of accessing MiMo-V2-Pro, at $2 per million tokens for input with up to 1 million tokens of context, or $1 per million with a 256,000 token limit, represents a significant advantage.
While ChatGPT remains a household name, the emergence of MiMo-V2-Pro demonstrates that innovation doesn't always require exorbitant budgets. Xiaomi has proven it can build a potent AI model and sell it at a fraction of the price. The real story here isn't just about a new language model; it's about the potential for a Chinese tech giant to reshape the landscape of accessible AI.