TellWorldAI World Archive

Home›Models

Model

GLM-4.5 and GLM-4.5-Air

GLM-4.5 is Zhipu's open-weight language model, released in July 2025 with 355 billion total parameters and a lighter sibling, GLM-4.5-Air. Both are hybrid reasoning models released under the MIT License.

Status AvailableModality TextOpen weights YesContext (K tokens) 128 Era Generative AI for everyone

Parameters
355B
License
MIT
Access
API, App, Download
Release date
Developer
Zhipu AI (Z.ai)

What it is

GLM-4.5 is a large language model from Zhipu (Z.ai, 智谱), released in July 2025 together with GLM-4.5-Air, a smaller model in the same family. The weights were posted to Hugging Face on 20 July 2025.[2][3]

Origins

The release came after a run of earlier models. In April 2025 Zhipu had released the GLM-4-0414 and GLM-Z1 reasoning models under the MIT License.[2][3]

Main features

GLM-4.5 has 355 billion total parameters, of which 32 billion are active. GLM-4.5-Air has 106 billion total parameters and 12 billion active.[1] Total parameters are all the weights in a model; active parameters are the ones used to process a given input. A gap between the two numbers reflects a sparse design such as the mixture-of-experts (MoE) architecture, in which only part of the network runs for each token.

Both models are hybrid reasoning models. Each has a thinking mode for complex reasoning and tool use, and a non-thinking mode for instant answers.[1] Tool use means the model can call external functions, such as search or code execution, while it works on a task.

The models support up to 128K tokens of context and OpenAI-style tool calling. They are served through the Z.ai API Platform and the Zhipu AI Open Platform.[1] A context window is the amount of text a model can take into account at one time.

Across 12 industry benchmarks, GLM-4.5 scored 63.2, ranking third among all proprietary and open models. GLM-4.5-Air scored 59.8.[1] A benchmark is a standard test set used to compare models on the same tasks.

Wikipedia notes that GLM-4.5 can run on eight NVIDIA H20 chips.[2] The technical report, arXiv 2508.06471, was published on 8 August 2025.[1]

Licence

Both models are released under the MIT License, which permits commercial use. Wikipedia says the GLM family has been MIT-licensed since July 2025.[1][2] MIT is a short, permissive open-source licence, so users can use, modify and redistribute the models with few restrictions.

Its place in development

GLM-4.5V followed in August 2025, and GLM-4.6 in September 2025.[2] GLM-4.5 is therefore the model that sits between the April 2025 GLM-4-0414 and GLM-Z1 releases and the later versions of the series.

Version history

7
  1. 2025
    • Jul 2025 GLM-4.5 and GLM-4.5-Air · This page

      GLM-4.5 is Zhipu's open-weight language model, released in July 2025 with 355 billion total parameters and a lighter sibling, GLM-4.5-Air. Both are hybrid reasoning models released under the MIT License.

    • Sep 2025 GLM-4.6

      GLM-4.6 is an open-weights large language model released by Zhipu in late September 2025, with a 200K-token context window, about 357 billion parameters and an MIT licence, aimed at coding, reasoning and agent work.

    • Dec 2025 GLM-4.7

      GLM-4.7 is Zhipu's open-weights language model released in December 2025, aimed at multilingual agentic coding and terminal tasks. It reports clear gains over GLM-4.6 on software-engineering benchmarks and ships under the MIT license.

  2. 2026
    • Feb 12, 2026 GLM-5

      GLM-5 is Zhipu's open model released on 12 February 2026 under the MIT License, with 744 billion parameters and 40 billion active. It was built for agentic engineering.

    • Apr 7, 2026 GLM-5.1

      GLM-5.1 is Zhipu AI's open-source large language model released in April 2026, built for coding and long-running agent tasks and notable for topping SWE-Bench Pro ahead of competing frontier models in Zhipu's own comparison.

    • Jun 16, 2026 GLM-5.2

      Zhipu's open-weight language model released in June 2026 under the MIT License, with a 1M-token context, a shared sparse-attention indexer, and higher reasoning and coding scores than GLM-5.1.

    • Aug 14, 2026 GLM-5.3

      GLM-5.3 is Zhipu's open-weight language model released on 14 August 2026. It keeps the GLM-5.2 base model and gains its improvements from post-training, with reported state-of-the-art open results on agentic coding and cybersecurity benchmarks.

20251

  1. ReleaseZhipu releases GLM-4.5 for agentic tasks under the MIT License

    In July 2025 Zhipu released GLM-4.5 and GLM-4.5-Air under the MIT License. Both are hybrid reasoning models built for agentic tasks, and GLM-4.5 ranked third in a 12-benchmark comparison.

Developed by1

Zhipu AI (Z.ai)

Builds on1

ChatGLM-6B

Foundation for1

GLM-4.5V

Succeeded by1

GLM-4.6

Notes and sources

  1. zai-org/GLM-4.5https://huggingface.co/zai-org/GLM-4.5
  2. Z.ai (Zhipu AI)https://en.wikipedia.org/wiki/Zhipu_AI
  3. zai-org modelshttps://huggingface.co/api/models?author=zai-org&sort=createdAt

Cite this page: GLM-4.5 and GLM-4.5-Air. TellWorld, https://www.tellworld.net/models/glm-4-5. Updated 2026-11-08