Skip to content

GLM 5 VRAM requirements

How much VRAM GLM 5 needs
4-bit weights: 426 GBWith 8K context: 427 GBWith 32K context: 430 GB8-bit, 8K context: 748 GB
Parameters
754B
Active per token
42.4B
Longest context
198K
Released
Feb 2026
License
MIT
Strongest at
Not rated

How much VRAM GLM 5 needs

Memory GLM 5 needs at each precision and context length
PrecisionWeightsWith 8KWith 32KWith 128K
16-bit1,404 GB1,406 GB1,408 GB1,417 GB
8-bit746 GB748 GB750 GB759 GB
6-bit576 GB577 GB580 GB589 GB
5-bit499 GB501 GB503 GB512 GB
4-bit426 GB427 GB430 GB439 GB
3-bit343 GB345 GB347 GB356 GB
2-bit263 GB265 GB267 GB276 GB
1.6-bit145 GB147 GB149 GB158 GB

One conversation, 16-bit context cache, runtime overhead included.

GPUs and Macs that run GLM 5

The best way GLM 5 runs on common GPUs, Macs and AI boxes
MachineBest fitContextSpeedFirst replyReport
RTX 4060Too bigPDF: GLM 5 on RTX 4060 report
RTX 5060Too bigPDF: GLM 5 on RTX 5060 report
RTX 3070Too bigPDF: GLM 5 on RTX 3070 report
RTX 3080 10 GBToo bigPDF: GLM 5 on RTX 3080 10 GB report
RTX 3060 12 GBToo bigPDF: GLM 5 on RTX 3060 12 GB report
Arc B580Too bigPDF: GLM 5 on Arc B580 report
RTX 4070Too bigPDF: GLM 5 on RTX 4070 report
RTX 4070 SuperToo bigPDF: GLM 5 on RTX 4070 Super report
RTX 5070Too bigPDF: GLM 5 on RTX 5070 report
RTX 4060 Ti 16 GBToo bigPDF: GLM 5 on RTX 4060 Ti 16 GB report
RTX 5060 Ti 16 GBToo bigPDF: GLM 5 on RTX 5060 Ti 16 GB report
Arc A770 16 GBToo bigPDF: GLM 5 on Arc A770 16 GB report
Radeon RX 9070 XTToo bigPDF: GLM 5 on Radeon RX 9070 XT report
RTX 4070 Ti SuperToo bigPDF: GLM 5 on RTX 4070 Ti Super report
RTX 4080 SuperToo bigPDF: GLM 5 on RTX 4080 Super report
RTX 5070 TiToo bigPDF: GLM 5 on RTX 5070 Ti report
RTX 5080Too bigPDF: GLM 5 on RTX 5080 report
Radeon RX 7900 XTXToo bigPDF: GLM 5 on Radeon RX 7900 XTX report
RTX 3090Too bigPDF: GLM 5 on RTX 3090 report
RTX 4090Too bigPDF: GLM 5 on RTX 4090 report
Radeon AI PRO R9700Too bigPDF: GLM 5 on Radeon AI PRO R9700 report
RTX 5090Too bigPDF: GLM 5 on RTX 5090 report
2× RTX 3090Too bigPDF: GLM 5 on 2× RTX 3090 report
2× RTX 4090Too bigPDF: GLM 5 on 2× RTX 4090 report
RTX PRO 5000 Blackwell 48 GBToo bigPDF: GLM 5 on RTX PRO 5000 Blackwell 48 GB report
RTX 6000 AdaToo bigPDF: GLM 5 on RTX 6000 Ada report
L40SToo bigPDF: GLM 5 on L40S report
2× RTX 5090Too bigPDF: GLM 5 on 2× RTX 5090 report
4× RTX 3090Too bigPDF: GLM 5 on 4× RTX 3090 report
RTX PRO 6000 BlackwellToo bigPDF: GLM 5 on RTX PRO 6000 Blackwell report
Mac mini M4 16 GBToo bigPDF: GLM 5 on Mac mini M4 16 GB report
Mac mini M4 24 GBToo bigPDF: GLM 5 on Mac mini M4 24 GB report
Mac mini M4 32 GBToo bigPDF: GLM 5 on Mac mini M4 32 GB report
Mac mini M4 Pro 24 GBToo bigPDF: GLM 5 on Mac mini M4 Pro 24 GB report
Mac mini M4 Pro 48 GBToo bigPDF: GLM 5 on Mac mini M4 Pro 48 GB report
Mac mini M4 Pro 64 GBToo bigPDF: GLM 5 on Mac mini M4 Pro 64 GB report
M4 Max 128 GBToo bigPDF: GLM 5 on M4 Max 128 GB report
M5 Max 128 GBToo bigPDF: GLM 5 on M5 Max 128 GB report
M3 Ultra 96 GBToo bigPDF: GLM 5 on M3 Ultra 96 GB report
M3 Ultra 256 GB1.6-bit198K37 tok/s27 sPDF: GLM 5 on M3 Ultra 256 GB report
M3 Ultra 512 GB3-bit198K20 tok/s27 sPDF: GLM 5 on M3 Ultra 512 GB report
M5 Ultra 512 GB3-bit198K27 tok/s8.3 sPDF: GLM 5 on M5 Ultra 512 GB report
Ryzen AI Max+ 395 64 GBToo bigPDF: GLM 5 on Ryzen AI Max+ 395 64 GB report
Ryzen AI Max+ 395 128 GBToo bigPDF: GLM 5 on Ryzen AI Max+ 395 128 GB report
RTX SparkToo bigPDF: GLM 5 on RTX Spark report
DGX SparkToo bigPDF: GLM 5 on DGX Spark report
Dual DGX Spark1.6-bit198K20 tok/s5.4 sPDF: GLM 5 on Dual DGX Spark report
DGX Station GB3001.6-bit198K155 tok/s0.8 sPDF: GLM 5 on DGX Station GB300 report
2× RTX PRO 6000 Blackwell1.6-bit198K86 tok/s1.3 sPDF: GLM 5 on 2× RTX PRO 6000 Blackwell report
4× RTX PRO 6000 Blackwell3-bit192K47 tok/s0.7 sPDF: GLM 5 on 4× RTX PRO 6000 Blackwell report
8× RTX PRO 6000 Blackwell6-bit198K31 tok/s0.4 sPDF: GLM 5 on 8× RTX PRO 6000 Blackwell report
A100 80 GB SXMToo bigPDF: GLM 5 on A100 80 GB SXM report
H100 SXMToo bigPDF: GLM 5 on H100 SXM report
4× H200 NVL5-bit198K73 tok/s0.7 sPDF: GLM 5 on 4× H200 NVL report
8× H200 NVL8-bit198K55 tok/s0.4 sPDF: GLM 5 on 8× H200 NVL report

Best fit is the highest quality, up to 8-bit, that leaves 16K of context. First reply is the wait for a 2K prompt. Speeds are rough, plus or minus 25%.

How to run it

llama.cpp
llama-server -hf unsloth/GLM-5-GGUF
vLLM
vllm serve zai-org/GLM-5
MLX (Mac)
mlx_lm.chat --model mlx-community/GLM-5-4bit

Questions

How much VRAM does GLM 5 need?

GLM 5 needs 426 GB for its weights at 4-bit, 427 GB with an 8K context and 430 GB with 32K. At 8-bit, 748 GB.

What is the minimum VRAM for GLM 5?

About 147 GB at 1.6-bit with an 8K context, the lowest precision we recommend for a model this size. The smallest machine here that runs it without spilling is the M3 Ultra 256 GB.

How big is GLM 5?

754B parameters (42.4B active per token): 1,404 GB of weights at 16-bit, 426 GB at 4-bit.

Can GLM 5 run on an RTX 4090?

Not on one card. The smallest machine here that runs it is the M3 Ultra 256 GB.

More models