RUN THIS LLM
Search local LLM hardware requirements
GLM-4.5 Air 106B MoE
Zhipu · 106B MoE · General
106B MoE with 12B active params. Dual reasoning/non-thinking modes. Strong at coding, reasoning, and agentic tool use.
VRAM Requirements
| Quantization | VRAM |
|---|---|
| Q4_K_M (smallest) | 63.6 GB |
| Q8_0 (balanced) | 116.6 GB |
| FP16 (full quality) | 212 GB |
Specifications
- Parameters: 106B MoE (12B active per token)
- Category: General
- Max context: 128K tokens
- System RAM: 80 GB minimum
- HuggingFace: zai-org/GLM-4.5-Air
Loading interactive analysis...