GLM-4.5 Air
MoEToolsThinkingZ.ai · 106B (12B active) · Mixture of Experts
Consumer-friendly GLM MoE — 12B active, strong agentic & tool use
0
0
2025-07
128K tokens
Use cases
Your hardware
16 GB Reference Desktop
Quantization options
What is quantization?| Quant | Bits | VRAM | Disk | Quality | Status |
|---|---|---|---|---|---|
| Q2_K | 2 | 34.4 GB | 32.4 GB | low | Barely runs · ~1 t/s |
| Q3_K_M | 3 | 48 GB | 45.3 GB | moderate | Too heavy · ~1 t/s |
| Q4_K_M | 4 | 54.8 GB | 51.8 GB | good | Too heavy · ~1 t/s |
| Q5_K_M | 5 | 68.4 GB | 64.8 GB | good | Too heavy · ~1 t/s |
| Q6_K | 6 | 81.9 GB | 77.7 GB | excellent | Too heavy · ~1 t/s |
| Q8_0 | 8 | 109.1 GB | 103.7 GB | excellent | Too heavy · ~1 t/s |
| F16 | 16 | 217.7 GB | 207.3 GB | lossless | Too heavy · ~1 t/s |
About this model
GLM-4.5 Air (GLM-4.5 air) is a 106B sparse mixture-of-experts (MoE) model activating 12B active per token from Z.ai in the GLM family, designed primarily for chat, reasoning, code workflows. Native reasoning/thinking tokens expand output length per query, so reserve extra KV cache headroom. Supports structured tool calling; use Q4_K_M or higher when strict JSON schema adherence matters. Supports up to 128K context tokens; start at 4K–8K with Q8_0 KV cache on tight GPUs.
Hardware planning baselines start at 59.2 GB minimum memory and 98.7 GB recommended memory. At Q4_K_M, weights plus baseline runtime overhead use roughly 54.8 GB of RAM/VRAM and 51.8 GB of disk space. At 106B, GLM-4.5 Air belongs firmly in the workstation and multi-GPU tier. Running it comfortably requires 98.7 GB of total memory across dual RTX GPUs, high-memory Apple Silicon (Mac Studio Ultra/Max), or MoE hybrid CPU-RAM offloading—mobile devices and standard 16 GB laptops are too constrained for practical speeds.
Deploy GLM-4.5 Air locally via llama.cpp with Flash Attention enabled. Under MIT License, review official weight cards before commercial redistribution.
Highlights
- Local Fit Baseline59.2 GB minimum and 98.7 GB recommended RAM/VRAM.
- Primary QuantQ4_K_M targets 54.8 GB memory and 51.8 GB disk.
- Context & Capabilities128K max context · Sparse MoE · Tool calling · Reasoning.
System requirements by device
Memory planning targets 59.2 GB minimum RAM/VRAM and 98.7 GB recommended memory. The baseline quant is Q4_K_M, estimated at 54.8 GB memory and 51.8 GB disk. Actual runtime usage shifts with context length, KV cache precision, and OS background overhead.
| Device | Notes |
|---|---|
| Mac (Apple silicon) | Requires a 48 GB–192 GB Mac Studio or MacBook Pro Max/Ultra with enough unified wired budget for 54.8 GB. |
| Windows / Linux GPU | Use dual/multi-GPU CUDA splitting (--tensor-split) or MoE CPU expert offloading (-ot) paired with 32–128 GB DDR5 system RAM. |
| Phones & Compact Edge Boards | Not practical on phones (59.2 GB min RAM exceeds Android LMK and iOS Jetsam limits). Pick a 2B–4B model on mobile. |
Start with a short session and leave memory for the operating system. A file fitting on disk does not guarantee that it will load in memory.
Estimated memory by quantization
These are rough estimates, not tested minimum requirements. Check the actual file and runtime documentation; context, offloading and operating-system overhead can change the memory needed.
| Quant Tier | Estimated memory | Disk Size | Quality |
|---|---|---|---|
| Q2_K | 34.4 GB | 32.4 GB | low |
| Q3_K_M | 48.0 GB | 45.3 GB | moderate |
| Q4_K_M | 54.8 GB | 51.8 GB | good |
| Q5_K_M | 68.4 GB | 64.8 GB | good |
| Q6_K | 81.9 GB | 77.7 GB | excellent |
| Q8_0 | 109.1 GB | 103.7 GB | excellent |
| F16 | 217.7 GB | 207.3 GB | lossless |
How to run GLM-4.5 Air on your device
Start with a short prompt and the smallest quant that still gives the quality you need. A clean first run is more useful than loading a larger file with no memory headroom.
Run GLM-4.5 Air on Multi-GPU or High-RAM Desktop
- Plan for 59.2 GB minimum combined memory and 98.7 GB recommended memory for Q4_K_M.
- Use llama.cpp `-ot ".ffn_.*_exps.=CPU"` or `--tensor-split` across multiple GPUs so attention and KV cache remain in fast VRAM.
- Enable Flash Attention (`-fa on`) and Q8_0 KV cache (`-ctk q8_0 -ctv q8_0`) to prevent context buffers from exceeding GPU headroom.
- Avoid attempting to run workstation-scale weights on smartphones or 8 GB laptops; use a 2B–8B checkpoint on portable devices.
Run GLM-4.5 Air on High-Memory Apple Silicon Mac
- Verify you are on a 48 GB–192 GB Mac Studio or MacBook Pro Max/Ultra capable of holding 54.8 GB in wired unified memory.
- Launch via `llama-server` or LM Studio with MLX/Metal enabled and set initial context conservatively to 4K–8K.
- Monitor Activity Monitor Memory Pressure to confirm zero SSD swap thrashing during token generation.
- Step down to a lighter quantization (e.g., Q3_K_M or IQ4_XS) if total memory utilization exceeds 75%.
GLM-4.5 Air vs alternatives
| Model | Params | Context | Min RAM | License |
|---|---|---|---|---|
| GLM-4.5 Air | 106B (12B active) | 128K | 59.2 GB | MIT |
| GLM-4 9B | 9B | 128K | 5 GB | GLM-4 |
| GLM-4.6 | 357B (32B active) | 195K | 199.5 GB | MIT |
| GLM-5 | 744B (40B active) | 128K | 415.7 GB | MIT |
| GLM-5.2 | 753B (40B active) | 1M | 420.8 GB | MIT |
| GLM-5.3 | 753B (40B active) | 1M | 420.8 GB | MIT |
Use this table to choose the smallest nearby model that still matches the job. A related model with lower RAM can be the better local choice when the current page fits only at the edge of available memory.
Frequently asked questions
- What is GLM-4.5 Air?
- GLM-4.5 Air is a 106B sparse mixture-of-experts model (12B active active per token) from Z.ai, supporting up to 128K tokens with 59.2 GB minimum RAM and 98.7 GB recommended RAM.
- How do I download and install GLM-4.5 air?
- Download a GGUF build for GLM-4.5 Air, start with Q4_K_M, and confirm the model fits locally before increasing context or quant quality.
- How much VRAM or RAM does GLM-4.5 Air require?
- At Q4_K_M, GLM-4.5 Air occupies roughly 54.8 GB of active memory and 51.8 GB on disk. Add 2–6 GB of spare headroom for KV cache buffers, runtime scratch space, and OS apps (98.7 GB recommended).
- What is the best way to run GLM-4.5 Air locally?
- Run GLM-4.5 Air in llama.cpp starting with Q4_K_M. Keep initial context at 4K–8K tokens and confirm 100% GPU/Metal offload before scaling context length.
- Can GLM-4.5 Air run on a consumer laptop or smartphone?
- No smartphone or 8–16 GB laptop can hold GLM-4.5 Air cleanly in local memory. It targets multi-GPU NVIDIA workstations, high-RAM Mac Studios, or MoE CPU-RAM hybrid rigs with at least 59.2 GB total memory.
- How does GLM-4.5 Air compare to GLM-4 9B?
- GLM-4 9B (9B, 5 GB min RAM) sits closest in size. Compare active parameters, context ceiling (128K vs 128K), and license terms (MIT) before switching.