According to CoinMeta, China’s Z.AI recently launched a new AI model, aiming to compete with Anthropic and OpenAI in the field of coding. GLM-5.3 scored 31.4% in the internal coding tests of Z.AI, surpassing Claude Opus’s 29.5% by 4.8%, while using fewer tokens. Claude Fable still leads with 39.5%. In addition, GLM-5.3 outperformed Anthropic and OpenAI in 3 out of 9 benchmark tests, including automationbench, gdpval-aa, v2, and cybergym. Since January IPO, the stock price of Z.AI has risen by 1000%, although it has fallen by about 57% since its historical high in June.