Xiaomi slashes AI model API prices by 99% in price war escalation

Xiaomi Corp. permanently reduced prices for its MiMo-V2.5 series large language model APIs by as much as 99%, effective globally from Wednesday, intensifying a price war among Chinese AI providers that has already seen DeepSeek slash flagship model costs by 75%.

"The new pricing removes the distinction based on context window length and makes AI inference affordable at scale for enterprise workloads," the company said in a statement. Xiaomi also introduced a Credits-based billing system under its optimized Token Plan, where usage increased five to eight times the original level without additional charges.

The MiMo-V2.5 Pro version now costs RMB0.025 per million tokens for input cache hits, a 99% reduction from its previous pricing, while the standard MiMo-V2.5 dropped to RMB0.02 per million tokens, a 98% cut. Output pricing fell to RMB6 per million tokens for the Pro version (down 86%) and RMB2 per million tokens for the standard version (down 93%). The adjustments apply globally.

The cuts place Xiaomi's inference pricing below DeepSeek's recently discounted V4-Pro, which charges $0.003625 per million tokens for cache-hit input — roughly equivalent to Xiaomi's RMB0.025 rate at current exchange rates — and up to $0.87 per million output tokens. DeepSeek made its 75% reduction permanent in May, citing efficiency gains rather than promotional discounts. "It is not a discount. It is an efficiency gain being passed through," Sanchit Vir Gogia, chief analyst at Greyhound Research, said of DeepSeek's move.

China's generative AI market is projected to grow more than 35% annually through the decade, according to government-backed industry estimates, with more than 300 registered generative AI services as of early 2026. The price war pits Xiaomi against DeepSeek, Alibaba's Tongyi Qianwen, Baidu's Ernie Bot, and Tencent's Hunyuan models, all of which have expanded large language model offerings over the past two years. Industry analysts said sustained price reductions could trigger competition similar to the cloud-computing pricing battles that reshaped China's internet sector during the 2010s.

Lower inference costs may also increase demand for computing infrastructure as companies deploy AI at scale. China's AI server and data-center markets are projected to grow at double-digit rates through the decade, benefiting domestic hardware suppliers including Huawei and Chinese server manufacturers seeking alternatives to restricted US technologies. The International Energy Agency has warned that global data-center electricity consumption could more than double to about 1,000 terawatt-hours annually by 2030 as AI adoption accelerates.

For Xiaomi, the aggressive pricing strategy could boost adoption of its AI services across its smartphone and IoT ecosystem, though the margin impact remains unclear. Xiaomi shares traded 1.8% lower in Hong Kong on Wednesday, with short selling accounting for 29.5% of turnover. The company did not disclose the expected revenue impact of the price cuts or the number of enterprise customers currently using its MiMo models.

This article is for informational purposes only and does not constitute investment advice.