U.S. Tech Giants Quietly Adopt Chinese Large Models Despite High Costs

With the escalating costs of premium U.S. large language models, a growing number of American tech firms are quietly embedding Chinese open-source AI into their production environments. Cryptocurrency exchange Coinbase recently led this shift, successfully slashing AI expenses by swapping models while maintaining full access for staff.
Cost Efficiency Takes Center Stage
Coinbase CEO Brian Armstrong announced that Zhipu GLM 5.2 and Moonshot Kimi K2.7 are now the default models for all engineers, primarily handling routine tasks like code reviews and document summaries. Despite an exponential rise in token usage, this strategy has reduced the company’s AI spending by nearly 50%.
Since most engineers had not hit previous usage caps, this optimization preserved their token quotas while demonstrating the approach’s scalability. Beyond Coinbase, major firms like Airbnb and AI startup Lindy have also migrated to various Chinese open-source models to manage rising costs.
Chinese Open-Source Models Gain Traction Abroad
Industry data indicates that several Chinese open-source models deliver performance on par with top-tier overseas alternatives at a fraction of the cost. On OpenRouter, a leading AI model API platform, Chinese models have consistently ranked among the most frequently used text models.
Currently, models such as DeepSeek, Xiaomi MiMo, MiniMax, Tencent Huan Yuan, and Zhipu GLM are gaining international favor due to their superior cost-performance ratio. This combination of technical capability and affordability is quietly reshaping the AI infrastructure strategies of global technology companies.
Related article
How to optimize SEO for Google Japan and Yahoo Japan?
Master Loom: Create and Share Professional Videos with AITable of ContentsIntroduction to LoomAccessing LoomRecording OptionsFull ScreenWindowCameraAudio and Sound QualityEditing and CustomizationBackground OptionsEditing the VideoAdding EffectsShari
How does Google Japan algorithm work for SEO?
AI-Generated MrBeast ThumbnailsTable of ContentsIntroductionMr Beast's Body Transformation ChallengeExploring Thumbnail Concepts for the ChallengeEnhancing Mr Beast's Body in PhotosApplying AI Techniques for Body TransformationsCreating Concept Thumb
Aliyun Qwen Hits 30M Users in First Month, Open-Source Context Drives New Enterprise Agent Paradigm
Just one month after its launch, Alibaba’s AI office suite "QwenWork" has achieved remarkable traction, surpassing 30 million users with over half being enterprise clients, signaling a significant uptick in corporate AI adoption.In the last 30 days,
Related Special Topic Recommendations
Comments (0)
0/500

With the escalating costs of premium U.S. large language models, a growing number of American tech firms are quietly embedding Chinese open-source AI into their production environments. Cryptocurrency exchange Coinbase recently led this shift, successfully slashing AI expenses by swapping models while maintaining full access for staff.
Cost Efficiency Takes Center Stage
Coinbase CEO Brian Armstrong announced that Zhipu GLM 5.2 and Moonshot Kimi K2.7 are now the default models for all engineers, primarily handling routine tasks like code reviews and document summaries. Despite an exponential rise in token usage, this strategy has reduced the company’s AI spending by nearly 50%.
Since most engineers had not hit previous usage caps, this optimization preserved their token quotas while demonstrating the approach’s scalability. Beyond Coinbase, major firms like Airbnb and AI startup Lindy have also migrated to various Chinese open-source models to manage rising costs.
Chinese Open-Source Models Gain Traction Abroad
Industry data indicates that several Chinese open-source models deliver performance on par with top-tier overseas alternatives at a fraction of the cost. On OpenRouter, a leading AI model API platform, Chinese models have consistently ranked among the most frequently used text models.
Currently, models such as DeepSeek, Xiaomi MiMo, MiniMax, Tencent Huan Yuan, and Zhipu GLM are gaining international favor due to their superior cost-performance ratio. This combination of technical capability and affordability is quietly reshaping the AI infrastructure strategies of global technology companies.
How to optimize SEO for Google Japan and Yahoo Japan?
Master Loom: Create and Share Professional Videos with AITable of ContentsIntroduction to LoomAccessing LoomRecording OptionsFull ScreenWindowCameraAudio and Sound QualityEditing and CustomizationBackground OptionsEditing the VideoAdding EffectsShari
How does Google Japan algorithm work for SEO?
AI-Generated MrBeast ThumbnailsTable of ContentsIntroductionMr Beast's Body Transformation ChallengeExploring Thumbnail Concepts for the ChallengeEnhancing Mr Beast's Body in PhotosApplying AI Techniques for Body TransformationsCreating Concept Thumb
Aliyun Qwen Hits 30M Users in First Month, Open-Source Context Drives New Enterprise Agent Paradigm
Just one month after its launch, Alibaba’s AI office suite "QwenWork" has achieved remarkable traction, surpassing 30 million users with over half being enterprise clients, signaling a significant uptick in corporate AI adoption.In the last 30 days,





Home






