Z.ai
Z.ai is a large language model API service platform operated by Zhipu AI, providing API access to the GLM series of models. The platform's interface design is highly compatible with the OpenAI API format, facilitating integration for developers familiar with that standard.
Core Facts
@ai-sdk/zai is the official Provider in the Vercel AI SDK for Zhipu AI models.
@ai-sdk/zai depends on @ai-sdk/openai-compatible, indicating that Zhipu GLM's API interface is largely compatible with OpenAI's calling conventions.
The models tested include Google's Gemini 3.7 Flash, Z.ai's GLM-5, Alibaba's Qwen 3 235B A22B Instruct, and OpenAI's gpt-oss-120b, all running on Kaggle.
Timeline (last 90 days)
GLM Coding Plan is a monthly subscription service launched by Z.ai (the international brand of Zhipu AI), used to call GLM models with a fixed quota in AI coding tools, rather than being billed by token.
Z.ai charges $1.40 per million tokens for fresh input to GLM-5.3 and $0.26 per million tokens for cached input, a difference of about 5.4 times.
The GLM Coding Plan subscription terms explicitly prohibit users from calling the model from their own apps, bots, websites, SaaS products, or other systems, the quota cannot be shared with colleagues, and it is non-refundable.
Z.ai's peak hours are Monday to Friday 14:00 to 18:00 Singapore time (UTC+8), totaling 20 hours per week; outside peak hours, model usage is billed at 50% of the standard rate.
The credit formula for GLM Coding Plan is: credits = (input tokens × input multiplier + cached input × cache multiplier + output × output multiplier) / 10,000.
The monthly prices in Z.ai's April 2026 migration notice were $18, $72 and $160 respectively, and Pro and Max rose by about 11% and 5% respectively when the credit system launched.
GLM Coding Plan fully switched to a token-metered credit system on July 30, 2026, after previously evolving from "counting by prompt, with no weekly cap" to "counting by prompt, with 5-hour and weekly caps".
Z.ai is an international product line (overseas brand) launched in 2024 by Zhipu AI (the Chinese large-model company Zhipu), and its GLM series models perform close to the international frontier in multiple benchmarks.
Anthropic recently released a report stating that a powerful new model released by the Chinese company Z.ai may pose a major cybersecurity threat.
GLM-Legal is built on the base model GLM-5.3-Flash, belonging to the 'base + domain fine-tuning' route.
23 more timeline events
All Facts (20)
@ai-sdk/zai is the official Provider in the Vercel AI SDK for Zhipu AI models.
90%Verified@ai-sdk/zai depends on @ai-sdk/openai-compatible, indicating that Zhipu GLM's API interface is largely compatible with OpenAI's calling conventions.
85%UnverifiedThe API service brand offered externally by Zhipu AI is Z.ai (formerly the BigModel open platform).
70%UnverifiedGLM-5.3-Flash, from Z.ai, is a multimodal MoE model with a total of 320B (320 billion) parameters, with only 18B parameters activated at any given time.
70%Partially VerifiedThe models tested include Google's Gemini 3.7 Flash, Z.ai's GLM-5, Alibaba's Qwen 3 235B A22B Instruct, and OpenAI's gpt-oss-120b, all running on Kaggle.
65%UnverifiedThe GLM Coding Plan subscription terms explicitly prohibit users from calling the model from their own apps, bots, websites, SaaS products, or other systems, the quota cannot be shared with colleagues, and it is non-refundable.
50%UnverifiedZ.ai's peak hours are Monday to Friday 14:00 to 18:00 Singapore time (UTC+8), totaling 20 hours per week; outside peak hours, model usage is billed at 50% of the standard rate.
50%UnverifiedThe credit formula for GLM Coding Plan is: credits = (input tokens × input multiplier + cached input × cache multiplier + output × output multiplier) / 10,000.
50%UnverifiedThe monthly prices in Z.ai's April 2026 migration notice were $18, $72 and $160 respectively, and Pro and Max rose by about 11% and 5% respectively when the credit system launched.
50%UnverifiedZ.ai charges $1.40 per million tokens for fresh input to GLM-5.3 and $0.26 per million tokens for cached input, a difference of about 5.4 times.
50%UnverifiedGLM Coding Plan is a monthly subscription service launched by Z.ai (the international brand of Zhipu AI), used to call GLM models with a fixed quota in AI coding tools, rather than being billed by token.
50%UnverifiedZ.ai is an international product line (overseas brand) launched in 2024 by Zhipu AI (the Chinese large-model company Zhipu), and its GLM series models perform close to the international frontier in multiple benchmarks.
50%UnverifiedGLM-Legal is built on the base model GLM-5.3-Flash, belonging to the 'base + domain fine-tuning' route.
50%UnverifiedCurrently, access to GLM-Legal is by application, rather than a directly open API.
50%UnverifiedThe GLM-Legal model launched by Z.ai achieved a 19.2% all-pass rate on the Harvey LAB legal Agent benchmark, surpassing Claude Fable 5 and Kimi K3.
50%UnverifiedGLM-5.3-Flash 预训练使用了30万亿token的多模态语料，是首个原生多模态的 GLM-5，支持文本、图像、视频和文件。
50%UnverifiedGLM-5.3-Flash 是第一个将稀疏注意力和线性注意力融合进单一混合系统的开放前沿模型。
50%Unverified根据 vLLM 配置和 Z.ai 文档，GLM-5.3-Flash 的45层模型将 KDA 线性注意力层与 NoPE 稀疏 MLA 层交替堆叠。
50%UnverifiedZ.ai 声称 GLM-5.3-Flash 的注意力计算量减少3.0倍，KV缓存缩小4.4倍，从而支持1M token上下文窗口（提示加输出不超过1,048,576）。
50%UnverifiedGLM-5.3-Flash 在国产AI芯片上规模化运行，搭配基于 SGLang 的专用推理引擎，声称端到端服务增益达到3倍。
50%