DeepSeek’s pricing shock is rippling through China’s AI sector, pressuring rivals and unsettling cloud providers
Chinese artificial intelligence developers are rapidly overhauling their pricing strategies as DeepSeek’s ultra-cheap V4 models trigger a new round of competition, forcing rivals to rethink how they monetise AI in a cutthroat domestic market.
Smartphone and electric-vehicle maker Xiaomi has emerged as one of the latest companies to respond, slashing application programming interface (API) costs for its MiMo-V2.5 model by as much as 99 per cent from previous levels.
Usage of Xiaomi’s MiMo-V2.5 and MiMo-V2.5-Pro models surged following last week’s pricing announcement, with MiMo-V2.5 climbing to sixth place on US-based model marketplace OpenRouter. The model processed 1.7 trillion tokens in the seven days to Monday, representing growth of more than 999 per cent from the previous week.
Do you have questions about the biggest topics and trends from around the world? Get the answers with SCMP Knowledge, our new platform of curated content with explainers, FAQs, analyses and infographics brought to you by our award-winning team.
Other Chinese AI firms are experimenting with different approaches to monetisation.
AI unicorn MiniMax on Monday launched its next-generation flagship model, MiniMax M3, pairing token-based billing with subscription plans ranging from US$7.24 to US$69.28 per month.
The transition has not been without friction.
Following the introduction of the new pricing model, some users complained that token consumption was significantly higher than expected, causing monthly quotas to be exhausted more quickly. MiniMax later apologised and said existing users with unlimited weekly access would retain those benefits.
Chinese AI companies are continuing to experiment with how they charge for AI services, reflecting the challenge of balancing user growth with sustainable revenue.
China’s AI price war was “becoming more sophisticated”, said Poe Zhao, China tech analyst and founder of the Hello China Tech newsletter.
Companies are increasingly experimenting with different charging models rather than relying solely on across-the-board price cuts. For enterprise users running complex workloads, the effective cost of completing a task may matter more than headline token prices.
Chinese developers are introducing new subscription tiers, restructuring billing systems and rolling out cheaper services in an effort to attract users who remain cautious about spending on AI products.
The pricing pressure is also rippling through the broader technology ecosystem. Third-party infrastructure providers have been forced to slash access fees to prevent developers from migrating to lower-cost alternatives.
Tencent Cloud this week cut API prices for DeepSeek’s V4 models by as much as 97.5 per cent, following the discounts introduced by DeepSeek on its AI offerings, including the V4 family launched in April.
Intense competition among inference providers has become a key driver of falling costs. Unlike proprietary systems, open-weight models can be hosted by multiple vendors, creating fierce competition among cloud providers and infrastructure companies.
The economic advantage of open models was highlighted in a November working paper by Frank Nagle of the Massachusetts Institute of Technology and the Linux Foundation, alongside Daniel Yue of the Georgia Institute of Technology. Their research found that open models cost, on average, just 15.66 per cent as much as closed-source alternatives.
Chinese firms have become particularly aggressive on pricing.
The top three positions in the “Pricing: Cache Discount” ranking compiled by benchmarking firm Artificial Analysis are currently occupied by Chinese companies.
Lower pricing is also being viewed as a way to secure valuable user data.
He Baohong, chief engineer at the state-backed China Academy of Information and Communications Technology, said at a digital economy forum last week that data availability remained one of the biggest constraints on further advances in AI model performance.
However, the escalating pricing pressure is reverberating along the tech supply chain.
Shares in Chinese cloud computing and server-rental companies have weakened since DeepSeek announced permanent price cuts, reflecting concerns that falling inference costs could eventually reduce demand for outsourced computing power.
More Articles from SCMP
Francis Lui’s Hot Delight stamps credentials as serious Classic Series contender at Sha Tin
Misdiagnosis isn’t misconduct, lawyer for accused doctor tells medical watchdog
China urged to fast-track carrier fleet upgrades as Japan boosts strike power
Nailed it: 4 moisturising treatments for your nails and cuticles
This article originally appeared on the South China Morning Post (www.scmp.com), the leading news media reporting on China and Asia.
Copyright (c) 2026. South China Morning Post Publishers Ltd. All rights reserved.
