China’s AI companies are done being cheap, and the revenue numbers prove it

For the better part of a year, Chinese AI companies ran what amounted to a global clearance sale on intelligence. Models that rivaled OpenAI and Anthropic at a fraction of the cost flooded the market, gobbled up developer mindshare, and turned API pricing into a race to the bottom.

Now the strategy is flipping. Across the sector, Chinese AI firms are raising prices, rolling out premium tiers, and chasing enterprise contracts.

The price hike heard around the API market

DeepSeek announced a significant API price increase in early August 2026. The new rates, which took effect on August 17, introduced a peak and off-peak pricing structure. Some tiers saw increases of up to 12x.

DeepSeek is far from alone. Industry-wide, average Chinese large language model API prices rose sharply in Q2 2026. Input prices climbed 48% quarter-on-quarter, while output prices jumped 80%.

Even with these increases, Chinese models remain dramatically cheaper than their Western competitors. Pricing still sits at roughly 10 to 60 times less than what OpenAI or Anthropic charge for comparable capabilities.

MiniMax and Z.ai show the enterprise playbook

MiniMax reported first-half 2026 revenue of $116.6 million, a 283% increase year-on-year. Enterprise services surged 703% to $73.9 million, accounting for 63.4% of total revenue. Annualized recurring revenue hit $800 million by August.

Z.ai is projected to approach nearly $1 billion in annual revenue run-rate, driven primarily by enterprise API deployments. Like MiniMax, Z.ai offers models via open-weight distribution. The commercial engine runs on paid API access and enterprise contracts rather than licensing the weights themselves.

Global usage tells the story

On platforms like OpenRouter, which aggregates access to dozens of AI models, Chinese providers captured between 30% and 46% of total token consumption. DeepSeek, Moonshot, and Z.ai frequently topped the usage rankings.

Profitability remains the open question

Revenue growth is one thing. Actually making money is another. Experts caution that profitability across the Chinese AI sector is still more aspiration than reality for most players, due to ongoing heavy infrastructure expenditure.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.