Open-weights AI model GLM-5.2 tops Intelligence Index with score of 51

Z AI's GLM-5.2 has claimed the top spot among open-weights models on the Artificial Analysis Intelligence Index, scoring 51 and surpassing rivals like MiniMax-M3 and DeepSeek V4 Pro, according to Hacker News. The model, which shares the same architecture as GLM-5.1 with 744 billion total parameters and 40 billion active parameters, improved by 11 points over its predecessor. "GLM-5.2 becomes top open weights AI model on Intelligence Index with score of 51" This leap in performance underscores Z AI's focus on advancing scientific reasoning and other evaluation benchmarks, as reported by eicker.news. The model is available through Z AI's first-party API and third-party providers, maintaining similar pricing to GLM-5.1 at $1.4 per 1 million input tokens, $4.4 per 1 million output tokens, and $0.26 per 1 million cache hit tokens, according to Hacker News. "The model shows notable improvements in scientific reasoning and other evaluations" With a score of 51, GLM-5.2 outperforms competitors MiniMax-M3 and DeepSeek V4 Pro, both of which scored 44, as noted by both outlets. This positions Z AI as a strong contender in the open-weights AI space, offering a powerful model at competitive pricing.

2 sources•19 JunRead the Baker's Take→