Chinese open-source AI models gain use without market reaction

Chinese companies released open-source AI models that rank among the most-used on public platforms. When DeepSeek released its R1 model in January 2025, U.S. tech company values fell by a reported trillion dollars and lawmakers proposed restrictions on government devices.

ai released GLM-5.2 last month, similar performance claims appeared but no comparable market or policy reaction followed.

U.S. firms typically host models in the cloud, charge for access, and limit visibility into model details. Open-source models from Chinese companies rank lower than the top closed models on some benchmarks yet perform adequately for many tasks. ai took 30 percent longer than a cloud model from OpenAI for the same task but incurred no cost.

GLM-5.2 ranked fifth on OpenRouter usage data, while DeepSeek’s latest model held the top position with more than twice the usage. Seven of the ten most-used models on that platform come from Chinese companies.

stability comparisons Artificial Analysis ranked GLM-5.2 as the most intelligent open-source model available. ai stated that GLM-5.2 outperforms OpenAI’s GPT-5.5 on a software engineering benchmark, though the same testing service found it slightly behind on intelligence tests.

Observers noted that Chinese open-source models can run on local servers and remain available even if U.S. models are withdrawn or altered.

market context The U.S. government has maintained export controls on advanced chips to China and has restricted foreign access to certain models at times. Commentators stated that continued restrictions could push development of separate technology ecosystems.

Business caution, compliance requirements, and existing enterprise relationships may limit adoption of downloaded open-source models inside large organizations. Europe lacks comparable AI development activity, according to one academic quoted by New Scientist, raising questions about long-term technological dependence.