GLM-5.3-Flash

智谱AI正式发布了GLM-5.3-Flash大模型，相关信息在Hacker News上引发关注1。根据Artificial Analysis的报告，该模型由Z AI于2026年8月26日推出2。GLM-5.3-Flash是一款闭源推理模型，采用扩展思维或思维链推理，且Z AI未披露其参数规模2。此外，该模型仅支持文本输入和输出，不支持图像输入，并具备400k token的上下文窗口2。

Artificial Analysis的分析报告显示，GLM-5.3-Flash在Artificial Analysis Intelligence Index上得分57，显著高于同价位推理模型18分的中位数2。在定价方面，其输入价格为0.15美元/百万token（同价位中位数为0.25美元），输出价格为0.50美元/百万token（同价位中位数为0.90美元）2。在相关评估中，该模型生成了150M个输出token（中位数为64M），评估总成本为138.02美元2。

Zhipu AI (Z AI) has released the GLM-5.3-Flash large language model, drawing significant attention on Hacker News 1. The model was released on August 26, 2026 2. It is a closed-source reasoning model that utilizes extended thinking or chain-of-thought reasoning 2. The model features a 400,000-token context window and supports only text input and output, without image input 2. Z AI has not disclosed the parameter size of the model 2.

Artificial Analysis published an analysis report on the model 2. GLM-5.3-Flash scored 57 on the Artificial Analysis Intelligence Index, significantly higher than the median of 18 for reasoning models at a similar price point 2. The pricing is $0.15 per million input tokens and $0.50 per million output tokens, compared to the respective medians of $0.25 and $0.90 2. During the evaluation, the model generated 150 million output tokens, surpassing the median of 64 million, and incurred a total evaluation cost of $138.02 2. The Hacker News discussion includes links to the official blog and the discussion page 1.