DistantNews
Support us
US developers ditch Anthropic for cheaper Chinese AI: 'Cost difference is not noticeable at all.' [Deep Dive] | Dong-A I
๐Ÿ‡ฐ๐Ÿ‡ท South Korea /Technology

US developers ditch Anthropic for cheaper Chinese AI: 'Cost difference is not noticeable at all.' [Deep Dive] | Dong-A Ilbo

From Dong-A Ilbo · () Korean

Translated from Korean, summarized and contextualized by DistantNews.

At a glance

News Named sources Context piece
  • Chinese AI model Zhipu AI's new model GLM 5.2 offers impressive performance at a fraction of the cost of US competitors.
  • GLM 5.2 ranks fourth in performance among AI models, surpassing Google's Gemini 3.5, and costs significantly less than models from Anthropic and OpenAI.
  • Despite security concerns, Chinese AI models are gaining significant market share among US developers due to their cost-effectiveness, with some US companies already adopting them.

The artificial intelligence landscape is shifting as Chinese AI model Zhipu AI's latest offering, GLM 5.2, garners attention for its remarkable cost-performance ratio. This new model has impressed US tech executives, with Guillermo Rauch, CEO of US company Vercel, stating on X, "I was genuinely impressed and almost shocked. This is going to be a game-changer." The model's performance places it fourth in AI analysis rankings, just behind top-tier models from Anthropic and OpenAI, and ahead of Google's Gemini 3.5. Crucially, GLM 5.2's API usage costs are approximately $4.4 per million output tokens, a mere fraction of the $50 charged by Anthropic's Claude 3 Opus or OpenAI's GPT-4 Turbo.

I was genuinely impressed and almost shocked. This is going to be a game-changer.

โ€” Guillermo RauchCEO of US company Vercel, commenting on Zhipu AI's GLM 5.2 model.

Adding to its appeal, GLM 5.2 is open-source, allowing anyone to download and run it on their own servers, a significant advantage for local AI development. This has led to a surge in Zhipu AI's stock price, nearly doubling in the ten days following the model's release, and rising over 1500% since its January listing in Hong Kong. Financial analysts like JP Morgan have revised revenue forecasts upward, predicting substantial growth.

Despite potential security concerns often associated with Chinese technology, data indicates a growing adoption of Chinese AI models outside of China. A report by Axios, citing data from AI model marketplace OpenRouter, shows that Chinese AI models' token consumption share has surged from below 10% in early 2025 to around 50% by June 2026. This trend is driven by the escalating costs of US AI models, forcing companies to seek more affordable alternatives. Developers are finding that the quality difference is often negligible for many tasks. Companies like Airbnb and Cursor have already acknowledged using Chinese AI models, and startups like Lindi have reported saving millions by switching from Anthropic to DeepSeek, another Chinese AI provider.

The cost difference is not noticeable at all.

โ€” Part-time developerAn American part-time developer who switched from Anthropic to DeepSeek, quoted in Rest of World.

Microsoft is even reportedly considering integrating Chinese DeepSeek models into its "Copilot for Microsoft 365" product to offer cheaper options. This evolving market dynamic pits the "closed, high-cost" US AI models against the "ultra-low-cost, open-source" Chinese alternatives. As the technological gap narrows, Chinese AI is rapidly closing in on US dominance, with American companies increasingly willing to embrace these cost-effective solutions.

There is no reason not to use it if you can get basic AI functions for a tenth of the price.

โ€” Flo CrivelloCEO of Lindi, a US AI agent startup, explaining the rationale for switching to Chinese AI models.
DistantNews Editorial

Originally published by Dong-A Ilbo in Korean. Translated, summarized, and contextualized by our editorial team with added local perspective. Read our editorial standards.