What is GLM?
GLM
GLM is a family of open-weight large language models developed by Z.ai (Zhipu AI), including versions such as GLM-5, GLM-5.1, and GLM-5.2. In the AI industry, GLM is discussed as a competitive open-weight alternative to closed frontier models, with analysts tracking its performance, cost, and impact on model adoption.
How it developed
- Jan 2026 - Nathan Lambert said GLM 4.7 was one of the first models that could replace a lot of his day-to-day work.
- Jul 2026 - Jay V noted that DeepSeek’s usage dipped around the time GLM came out but seemed to bounce back.
- Jul 2026 - Nathan Lambert commented that US-based companies are relying on lesser models because the closed frontier is inaccessible, implying GLM’s role as an accessible alternative.
- Jul 2026 - Nathaniel Whittemore said that Harvey and Fireworks pair an open-weight GLM worker with an Opus advisor for legal tasks, seeing improved performance over Opus alone for a fraction of the cost.
- Aug 2026 - Patrick O’Shaughnessy said that open-source models like GLM are not free because you still have to run inference, and the cost per answer can be significant.
In the evidence
Every line below is attributed to a named speaker.
Nathan Lambert on closed-model inaccessibility forcing US companies onto weaker open alternatives during a security incident.
“They had to use a worse model to defend themselves or to analyze the data, which is a horrible state to be in that we have US-based companies now relying on lesser models because the closed frontier is inaccessible to them.”Nathan Lambert · 22 Jul 2026
Harvey and Fireworks pair an open-weight GLM worker model with an Opus advisor for legal tasks, reportedly outperforming Opus alone at a fraction of the cost.
“Harvey and Fireworks pair an open weight GLM worker with an Opus advisor for legal tasks, seeing improved performance over Opus alone for a fraction of the cost.”Nathaniel Whittemore · 7 Jul 2026
Open-source AI models are not free: inference has real marginal costs, and some models like Kimmy are significantly more expensive per answer than standard alternatives.
“The open- source models the talk about them being free is bizarre to me because it's marginal costs, right? You still have to run inference like GLM or Kimmy. Kimmy is very expensive to serve. The cost per answer is significantly higher.”Patrick O'Shaughnessy · 18 Aug 2026
Despite Twitter chatter suggesting GLM is displacing DeepSeek, OpenCode usage data shows DeepSeek dips briefly at GLM launches but then bounces back.
“You can kind of see the DeepSeek one dip around the time GLM sort of comes out, but it's seems to sort of bounce back up after.”Jay V · 24 Jul 2026
GLM 4.7 is worth watching as one of the first open models capable of replacing day-to-day developer coding agent workflows.
“I think 4.7 from GLM was one of the first ones that could replace a lot of my day-to-day.”Nathan Lambert · 27 Jan 2026