What are the hardware requirements for GLM 4.5 Air?

GLM 4.5 Air requires approximately 209.8 GB of VRAM and supports a context window of 128K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed GLM 4.5 Air and how large is it?

GLM 4.5 Air is developed by zai-org, a model with 110.5b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

How does GLM 4.5 Air perform on standard benchmarks?

GLM 4.5 Air has the following published scores: LMSYS ELO 1373. Compare against reference models on this page or on the LLM Explorer leaderboards.

GLM 4.5 Air by zai-org — VRAM 209.8GB, 128K context

Name: GLM 4.5 Air
Author: zai-org

GLM 4.5 Air is an open-source language model by zai-org. Features: 110.5b LLM, VRAM: 209.8GB, Context: 128K, License: mit, LLM Explorer Score: 0.47, ELO: 1373.

Arxiv:2508.06471 Conversational Deploy:azure En Endpoints compatible Eval-results Glm4 moe Region:us Safetensors Sharded Tensorflow Zh

Model Card on HF 🤗: https://huggingface.co/zai-org/GLM-4.5-Air

GLM 4.5 Air Benchmarks

LMSys ELO 1373 vs 1272 (so35) 7.9%

^nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").

What is the LLM Explorer Rank (Score)

GLM 4.5 Air Parameters and Internals

LLM Name	GLM 4.5 Air
Repository 🤗	https://huggingface.co/zai-org/GLM-4.5-Air
Model Size	110.5b
Required VRAM	209.8 GB
Updated	2026-08-01
Maintainer	zai-org
Model Type	glm4_moe
Model Files	3.0 GB: 1-of-47 4.7 GB: 2-of-47 4.7 GB: 3-of-47 4.7 GB: 4-of-47 4.7 GB: 5-of-47 4.7 GB: 6-of-47 4.7 GB: 7-of-47 4.7 GB: 8-of-47 4.7 GB: 9-of-47 4.7 GB: 10-of-47 4.7 GB: 11-of-47 4.7 GB: 12-of-47 4.7 GB: 13-of-47 4.7 GB: 14-of-47 4.7 GB: 15-of-47 4.7 GB: 16-of-47 4.7 GB: 17-of-47 4.7 GB: 18-of-47 4.7 GB: 19-of-47 4.7 GB: 20-of-47 4.7 GB: 21-of-47 4.7 GB: 22-of-47 4.7 GB: 23-of-47 4.7 GB: 24-of-47 4.7 GB: 25-of-47 4.7 GB: 26-of-47 4.7 GB: 27-of-47 4.7 GB: 28-of-47 4.7 GB: 29-of-47 4.7 GB: 30-of-47 4.7 GB: 31-of-47 4.7 GB: 32-of-47 4.7 GB: 33-of-47 4.7 GB: 34-of-47 4.7 GB: 35-of-47 4.7 GB: 36-of-47 4.7 GB: 37-of-47 4.7 GB: 38-of-47 4.7 GB: 39-of-47 4.7 GB: 40-of-47 4.7 GB: 41-of-47 4.7 GB: 42-of-47 4.7 GB: 43-of-47 4.7 GB: 44-of-47 4.7 GB: 45-of-47
Supported Languages	en zh
Model Architecture	Glm4MoeForCausalLM
License	mit
Context Length	131072
Model Max Length	131072
Transformers Version	4.54.0
Vocabulary Size	151552
Torch Data Type	bfloat16

Rank the GLM 4.5 Air capabilities

Have you tried this model? Rate its performance — this feedback helps the ML community find the right model for their needs.

Instruction Following and Task Automation
Factuality and Completeness of Knowledge
Censorship and Alignment
Data Analysis and Insight Generation
Text Generation
Text Summarization and Feature Extraction
Code Generation
Multi-Language Support and Translation

Quantized Models of the GLM 4.5 Air

Model	Likes	Downloads	VRAM
GLM 4.5 Air 2bit	4	107	33 GB
GLM 4.5 Air AWQ 4bit	24	295843	63 GB
GLM 4.5 Air AWQ 4bit	29	13143	63 GB
GLM 4.5 Air GPTQ Int4 Int8Mix	10	13073	70 GB
GLM 4.5 Air AWQ	9	11345	61 GB
GLM 4.5 Air 8bit	11	2726	114 GB
GLM 4.5 Air AWQ FP16Mix	14	2185	73 GB
GLM 4.5 Air 4bit	28	1261	60 GB
GLM 4.5 Air 3bit	32	621	46 GB
GLM 4.5 Air 6bit	3	201	86 GB

Best Alternatives to GLM 4.5 Air

Best Alternatives	Context / RAM	Downloads	Likes
GLM 4.5 Air	128K / 209.8 GB	263343	8
GLM 4.5 Air Base	128K / 222.3 GB	594	46
GLM 4.5 Air Derestricted	128K / 209.8 GB	100	102
...5 Air Abliterated Lossytensors	128K / 221.7 GB	10	5