Zhipu AI Releases GLM-5, New Benchmark for Chinese Reasoning Models

%23%20Zhipu%20AI%20Releases%20GLM-5%0A%0AFlagship%20reasoning%20model%20launched%20same%20day%20as%20MiniMax%20M2.5.%0A%0A%23%23%20Performance%0A-%20MATH%3A%2083.5%25%20(%2B12pp%20vs%20GLM-4)%0A-%20HumanEval%3A%2089.2%25%0A-%20128K%20context%20window%0A%0A%23%23%20Context%0AZhipu%20listed%20on%20HKEX%20in%20January%202026.%20GLM-5%20is%20their%20first%20flagship%20as%20a%20public%20company.