# GLM-4.5-Air: hardware requirements
> GLM-4.5-Air (Z.ai) is a 106B (12B active) mixture-of-experts model with a 128K-token context window, released under the MIT license. It needs about 68GB of memory at Q4_K_M with 8K context.
GLM-4.5-Air is Z.ai's 106B mixture-of-experts with 12B active parameters, built for agentic coding and tool use with a switchable thinking mode. Z.ai's later releases went either much larger (the GLM-5 series) or smaller (GLM-4.7-Flash), so Air is still the GLM model aimed at 96GB to 128GB machines. It needs roughly 70GB at Q4_K_M with 8K context: a Mac or Strix Halo system with 96GB or more runs it fully in memory, and a 24GB GPU with 64GB of system RAM can run it with experts offloaded. Its full-attention KV cache is large, so long contexts cost more memory than on gpt-oss-120b or Qwen3.5 122B-A10B.

## Recommended hardware
- **Minimum GPU:** RTX PRO 5000 72GB (72GB), Q4_K_M, ~73 tokens/sec
- **Best value:** RTX PRO 6000 (96GB), Q5_K_M, ~80 tokens/sec
- **Fastest single GPU:** RTX PRO 6000 (96GB), Q5_K_M, ~80 tokens/sec
- **Multi-GPU:** 3× RTX 3090 (72GB), Q4_K_M, ~38 tokens/sec
- **Multi-GPU:** 3× RTX 4090 (72GB), Q4_K_M, ~43 tokens/sec
- **Multi-GPU:** 3× RTX 5090 (96GB), Q4_K_M, ~60 tokens/sec
- **Mac / unified memory:** Ryzen AI Max+ 395 (128GB), Q5_K_M, ~13 tokens/sec

## Memory by quantization
| Quant | Weights | Total @ 8K ctx | Total @ 32K ctx |
| --- | --- | --- | --- |
| Q2_K | 39.8GB | 43.1GB | 47.7GB |
| Q3_K_M | 51.7GB | 55.4GB | 60.0GB |
| Q4_K_M | 64.3GB | 68.3GB | 73.0GB |
| Q5_K_M | 75.5GB | 79.9GB | 84.6GB |
| Q6_K | 87.4GB | 92.2GB | 96.8GB |
| Q8_0 | 112.6GB | 118.1GB | 122.8GB |
| BF16 | 212.0GB | 220.5GB | 225.1GB |

## GPUs that run it fully in memory (fastest first)
| GPU | Memory | Quant | Speed | Price |
| --- | --- | --- | --- | --- |
| RTX PRO 6000 | 96GB | Q5_K_M | ~80 tok/s | ~$16,000 |
| RTX PRO 5000 72GB | 72GB | Q4_K_M | ~73 tok/s | ~$9,100 |
| M5 Ultra 512GB | 512GB | Q8_0 | ~28 tok/s | n/a |
| M5 Ultra 256GB | 256GB | Q8_0 | ~28 tok/s | ~$9,499 |
| M5 Max 128GB | 128GB | Q5_K_M | ~25 tok/s | ~$5,100 |
| M4 Max 128GB | 128GB | Q5_K_M | ~23 tok/s | $3,499 MSRP |
| M3 Ultra 512GB | 512GB | Q8_0 | ~20 tok/s | $9,499 MSRP |
| Ryzen AI Max+ 395 | 128GB | Q5_K_M | ~13 tok/s | ~$3,300 |
| DGX Spark | 128GB | Q6_K | ~12 tok/s | ~$6,950 |
| Ryzen AI Max+ PRO 495 | 192GB | Q8_0 | ~9.3 tok/s | ~$6,800 |

## How to run it
- llama.cpp: `llama-server -hf unsloth/GLM-4.5-Air-GGUF:Q4_K_M -c 8192 -ngl 99`
- MLX (Mac): `mlx_lm.chat --model mlx-community/GLM-4.5-Air-4bit`
- Weights: https://huggingface.co/zai-org/GLM-4.5-Air

## Strengths
- Strong agentic coding and tool calling for its size
- Toggleable thinking mode
- 12B active parameters keep it responsive on unified-memory machines
- MIT license

## Weaknesses
- Needs roughly 70GB at Q4_K_M
- Large KV cache (184 KiB per token) limits long contexts
- Text only
- Now trails gpt-oss-120b and Qwen3.5 122B-A10B on most benchmarks
Full page: https://llamabuilds.ai/models/glm-4-5-air · Hardware guide: https://llamabuilds.ai/models/glm-4-5-air/requirements

---
Source: https://llamabuilds.ai/models/glm-4-5-air · Updated October 14, 2026 · Llama Builds. Speeds marked ~ are estimates from our bandwidth model (https://llamabuilds.ai/methodology); measured results are labeled as such.