Has China's new AI model become the safest path amid US bans?

Following the US government's latest export control directive and Anthropic's decision to take its new Claude Fable 5 model entirely offline for all users, certain limitations have emerged for enterprises and developers relying on proprietary AI infrastructure.

Against this backdrop, Chinese startup Z.ai (formerly Zhipu AI) has unveiled GLM-5.2, a model engineered for long-horizon autonomous coding and engineering tasks. According to the developers, the model has outperformed even GPT-5.5 on long-horizon coding benchmarks.

Whether this new open-weights system will prove reliable in practice and at what scale it will emerge as a credible alternative globally remains to be seen — real-world project experience will tell in the coming weeks.

👉 Read more on Pivot.uz

@pivotglobaal