back to local model checker

GLM-5

MoEToolsThinking

Zhipu AI · 744B (40B active) · Mixture of Experts

MoE with 256 experts, 40B active — frontier-class agentic coding

MIT
Downloads

0

Likes

0

Released

2026-02

Context

128K tokens

Use cases

chatreasoningcode
Total experts256
Active experts8
Active params40.0B

Your hardware

16 GB Reference Desktop

Quant
Q4_K_M
Memory
381.6 GB
Speed
--
Fit
Too heavy

Quantization options

What is quantization?
QuantBitsVRAMDiskQualityStatus
Q2_K2238.7 GB227.4 GBlowToo heavy · ~1 t/s
Q3_K_M3334 GB318.3 GBmoderateToo heavy · ~1 t/s
Q4_K_M4381.6 GB363.8 GBgoodToo heavy · ~1 t/s
Q5_K_M5476.9 GB454.7 GBgoodToo heavy · ~1 t/s
Q6_K6572.1 GB545.7 GBexcellentToo heavy · ~1 t/s
Q8_08762.7 GB727.5 GBexcellentToo heavy · ~1 t/s
F16161524.9 GB1455.1 GBlosslessToo heavy · ~1 t/s

About this model

GLM-5 (GLM-5) is a 744B sparse mixture-of-experts (MoE) model activating 40B active per token from Zhipu 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 415.7 GB minimum memory and 692.9 GB recommended memory. At Q4_K_M, weights plus baseline runtime overhead use roughly 381.6 GB of RAM/VRAM and 363.8 GB of disk space. At 744B, GLM-5 belongs firmly in the workstation and multi-GPU tier. Running it comfortably requires 692.9 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-5 locally via llama.cpp with Flash Attention enabled. Under MIT License, review official weight cards before commercial redistribution.

Highlights

  • Local Fit Baseline415.7 GB minimum and 692.9 GB recommended RAM/VRAM.
  • Primary QuantQ4_K_M targets 381.6 GB memory and 363.8 GB disk.
  • Context & Capabilities128K max context · Sparse MoE · Tool calling · Reasoning.

System requirements by device

Memory planning targets 415.7 GB minimum RAM/VRAM and 692.9 GB recommended memory. The baseline quant is Q4_K_M, estimated at 381.6 GB memory and 363.8 GB disk. Actual runtime usage shifts with context length, KV cache precision, and OS background overhead.

DeviceNotes
Mac (Apple silicon)Requires a 48 GB–192 GB Mac Studio or MacBook Pro Max/Ultra with enough unified wired budget for 381.6 GB.
Windows / Linux GPUUse dual/multi-GPU CUDA splitting (--tensor-split) or MoE CPU expert offloading (-ot) paired with 32–128 GB DDR5 system RAM.
Phones & Compact Edge BoardsNot practical on phones (415.7 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 TierEstimated memoryDisk SizeQuality
Q2_K238.7 GB227.4 GBlow
Q3_K_M334.0 GB318.3 GBmoderate
Q4_K_M381.6 GB363.8 GBgood
Q5_K_M476.9 GB454.7 GBgood
Q6_K572.1 GB545.7 GBexcellent
Q8_0762.7 GB727.5 GBexcellent
F161524.9 GB1455.1 GBlossless

How to run GLM-5 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.

Windows

Run GLM-5 on Multi-GPU or High-RAM Desktop

  1. Plan for 415.7 GB minimum combined memory and 692.9 GB recommended memory for Q4_K_M.
  2. Use llama.cpp `-ot ".ffn_.*_exps.=CPU"` or `--tensor-split` across multiple GPUs so attention and KV cache remain in fast VRAM.
  3. Enable Flash Attention (`-fa on`) and Q8_0 KV cache (`-ctk q8_0 -ctv q8_0`) to prevent context buffers from exceeding GPU headroom.
  4. Avoid attempting to run workstation-scale weights on smartphones or 8 GB laptops; use a 2B–8B checkpoint on portable devices.
Device how-to guide
Mac

Run GLM-5 on High-Memory Apple Silicon Mac

  1. Verify you are on a 48 GB–192 GB Mac Studio or MacBook Pro Max/Ultra capable of holding 381.6 GB in wired unified memory.
  2. Launch via `llama-server` or LM Studio with MLX/Metal enabled and set initial context conservatively to 4K–8K.
  3. Monitor Activity Monitor Memory Pressure to confirm zero SSD swap thrashing during token generation.
  4. Step down to a lighter quantization (e.g., Q3_K_M or IQ4_XS) if total memory utilization exceeds 75%.
Device how-to guide

GLM-5 vs alternatives

ModelParamsContextMin RAMLicense
GLM-5744B (40B active)128K415.7 GBMIT
GLM-5.2753B (40B active)1M420.8 GBMIT
GLM-5.3753B (40B active)1M420.8 GBMIT
GLM-5.1754B (40B active)128K421.3 GBMIT
GLM-4.6357B (32B active)195K199.5 GBMIT
GLM-4.5 Air106B (12B active)128K59.2 GBMIT

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-5?
GLM-5 is a 744B sparse mixture-of-experts model (40B active active per token) from Zhipu AI, supporting up to 128K tokens with 415.7 GB minimum RAM and 692.9 GB recommended RAM.
How do I download and install GLM-5?
Download a GGUF build for GLM-5, start with Q4_K_M, and confirm the model fits locally before increasing context or quant quality.
How much VRAM or RAM does GLM-5 require?
At Q4_K_M, GLM-5 occupies roughly 381.6 GB of active memory and 363.8 GB on disk. Add 2–6 GB of spare headroom for KV cache buffers, runtime scratch space, and OS apps (692.9 GB recommended).
What is the best way to run GLM-5 locally?
Run GLM-5 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-5 run on a consumer laptop or smartphone?
No smartphone or 8–16 GB laptop can hold GLM-5 cleanly in local memory. It targets multi-GPU NVIDIA workstations, high-RAM Mac Studios, or MoE CPU-RAM hybrid rigs with at least 415.7 GB total memory.
How does GLM-5 compare to GLM-5.2?
GLM-5.2 (753B, 420.8 GB min RAM) sits closest in size. Compare active parameters, context ceiling (128K vs 1M), and license terms (MIT) before switching.