Skip to content

GLM-4.7 Flash VRAM requirements

How much VRAM GLM-4.7 Flash needs
4-bit weights: 17.6 GBWith 8K context: 18.8 GBWith 32K context: 20.1 GB8-bit, 8K context: 32.1 GB
Parameters
31.2B
Active per token
3B
Longest context
198K
Released
Jan 2026
License
MIT
Strongest at
Chat, Coding, Agents

How much VRAM GLM-4.7 Flash needs

Memory GLM-4.7 Flash needs at each precision and context length
PrecisionWeightsWith 8KWith 32KWith 128K
16-bit58.2 GB59.4 GB60.6 GB65.6 GB
8-bit30.9 GB32.1 GB33.4 GB38.3 GB
6-bit23.8 GB25.1 GB26.3 GB31.3 GB
5-bit20.7 GB21.9 GB23.1 GB28.1 GB
4-bit17.6 GB18.8 GB20.1 GB25.0 GB
3-bit14.2 GB15.4 GB16.7 GB21.6 GB
2-bit10.9 GB12.1 GB13.4 GB18.3 GB
1.6-bit6.0 GB7.2 GB8.5 GB13.4 GB

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

GPUs and Macs that run GLM-4.7 Flash

The best way GLM-4.7 Flash runs on common GPUs, Macs and AI boxes
MachineBest fitContextSpeedFirst replyReport
RTX 4060Too bigPDF: GLM-4.7 Flash on RTX 4060 report
RTX 5060Too bigPDF: GLM-4.7 Flash on RTX 5060 report
RTX 3070Too bigPDF: GLM-4.7 Flash on RTX 3070 report
RTX 3080 10 GBToo bigPDF: GLM-4.7 Flash on RTX 3080 10 GB report
RTX 3060 12 GBToo bigPDF: GLM-4.7 Flash on RTX 3060 12 GB report
Arc B580Too bigPDF: GLM-4.7 Flash on Arc B580 report
RTX 4070Too bigPDF: GLM-4.7 Flash on RTX 4070 report
RTX 4070 SuperToo bigPDF: GLM-4.7 Flash on RTX 4070 Super report
RTX 5070Too bigPDF: GLM-4.7 Flash on RTX 5070 report
RTX 4060 Ti 16 GBToo bigPDF: GLM-4.7 Flash on RTX 4060 Ti 16 GB report
RTX 5060 Ti 16 GBToo bigPDF: GLM-4.7 Flash on RTX 5060 Ti 16 GB report
Arc A770 16 GBToo bigPDF: GLM-4.7 Flash on Arc A770 16 GB report
Radeon RX 9070 XTToo bigPDF: GLM-4.7 Flash on Radeon RX 9070 XT report
RTX 4070 Ti SuperToo bigPDF: GLM-4.7 Flash on RTX 4070 Ti Super report
RTX 4080 SuperToo bigPDF: GLM-4.7 Flash on RTX 4080 Super report
RTX 5070 TiToo bigPDF: GLM-4.7 Flash on RTX 5070 Ti report
RTX 5080Too bigPDF: GLM-4.7 Flash on RTX 5080 report
Radeon RX 7900 XTX5-bit24K126 tok/s0.8 sPDF: GLM-4.7 Flash on Radeon RX 7900 XTX report
RTX 30905-bit24K165 tok/s0.5 sPDF: GLM-4.7 Flash on RTX 3090 report
RTX 40905-bit24K171 tok/s0.2 sPDF: GLM-4.7 Flash on RTX 4090 report
Radeon AI PRO R97006-bit104K94 tok/s0.5 sPDF: GLM-4.7 Flash on Radeon AI PRO R9700 report
RTX 50906-bit104K209 tok/s0.2 sPDF: GLM-4.7 Flash on RTX 5090 report
2× RTX 30908-bit198K130 tok/s0.3 sPDF: GLM-4.7 Flash on 2× RTX 3090 report
2× RTX 40908-bit198K136 tok/s0.1 sPDF: GLM-4.7 Flash on 2× RTX 4090 report
RTX PRO 5000 Blackwell 48 GB8-bit198K161 tok/s0.3 sPDF: GLM-4.7 Flash on RTX PRO 5000 Blackwell 48 GB report
RTX 6000 Ada8-bit198K132 tok/s0.2 sPDF: GLM-4.7 Flash on RTX 6000 Ada report
L40S8-bit198K124 tok/s0.3 sPDF: GLM-4.7 Flash on L40S report
2× RTX 50908-bit198K187 tok/s0.1 sPDF: GLM-4.7 Flash on 2× RTX 5090 report
4× RTX 30908-bit198K130 tok/s0.2 sPDF: GLM-4.7 Flash on 4× RTX 3090 report
RTX PRO 6000 Blackwell8-bit198K187 tok/s0.1 sPDF: GLM-4.7 Flash on RTX PRO 6000 Blackwell report
Mac mini M4 16 GBToo bigPDF: GLM-4.7 Flash on Mac mini M4 16 GB report
Mac mini M4 24 GBToo bigPDF: GLM-4.7 Flash on Mac mini M4 24 GB report
Mac mini M4 32 GB4-bit36K40 tok/s12 sPDF: GLM-4.7 Flash on Mac mini M4 32 GB report
Mac mini M4 Pro 24 GBToo bigPDF: GLM-4.7 Flash on Mac mini M4 Pro 24 GB report
Mac mini M4 Pro 48 GB8-bit52K48 tok/s5.9 sPDF: GLM-4.7 Flash on Mac mini M4 Pro 48 GB report
Mac mini M4 Pro 64 GB8-bit198K48 tok/s5.9 sPDF: GLM-4.7 Flash on Mac mini M4 Pro 64 GB report
M4 Max 128 GB8-bit198K70 tok/s2.9 sPDF: GLM-4.7 Flash on M4 Max 128 GB report
M5 Max 128 GB8-bit198K74 tok/s0.8 sPDF: GLM-4.7 Flash on M5 Max 128 GB report
M3 Ultra 96 GB8-bit198K80 tok/s1.6 sPDF: GLM-4.7 Flash on M3 Ultra 96 GB report
M3 Ultra 256 GB8-bit198K80 tok/s1.6 sPDF: GLM-4.7 Flash on M3 Ultra 256 GB report
M3 Ultra 512 GB8-bit198K80 tok/s1.6 sPDF: GLM-4.7 Flash on M3 Ultra 512 GB report
M5 Ultra 512 GB8-bit198K94 tok/s0.5 sPDF: GLM-4.7 Flash on M5 Ultra 512 GB report
Ryzen AI Max+ 395 64 GB8-bit198K40 tok/s1.7 sPDF: GLM-4.7 Flash on Ryzen AI Max+ 395 64 GB report
Ryzen AI Max+ 395 128 GB8-bit198K40 tok/s1.7 sPDF: GLM-4.7 Flash on Ryzen AI Max+ 395 128 GB report
RTX Spark8-bit198K56 tok/s0.6 sPDF: GLM-4.7 Flash on RTX Spark report
DGX Spark8-bit198K51 tok/s0.7 sPDF: GLM-4.7 Flash on DGX Spark report
Dual DGX Spark8-bit198K51 tok/s0.4 sPDF: GLM-4.7 Flash on Dual DGX Spark report
DGX Station GB3008-bit198K289 tok/s0.1 sPDF: GLM-4.7 Flash on DGX Station GB300 report
2× RTX PRO 6000 Blackwell8-bit198K187 tok/s0.1 sPDF: GLM-4.7 Flash on 2× RTX PRO 6000 Blackwell report
4× RTX PRO 6000 Blackwell8-bit198K187 tok/s0.1 sPDF: GLM-4.7 Flash on 4× RTX PRO 6000 Blackwell report
8× RTX PRO 6000 Blackwell8-bit198K187 tok/s0.1 sPDF: GLM-4.7 Flash on 8× RTX PRO 6000 Blackwell report
A100 80 GB SXM8-bit198K198 tok/s0.4 sPDF: GLM-4.7 Flash on A100 80 GB SXM report
H100 SXM8-bit198K239 tok/s0.1 sPDF: GLM-4.7 Flash on H100 SXM report
4× H200 NVL8-bit198K265 tok/s0.1 sPDF: GLM-4.7 Flash on 4× H200 NVL report
8× H200 NVL8-bit198K265 tok/s0.1 sPDF: GLM-4.7 Flash 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

Ollama
ollama run glm-4.7-flash:latest
llama.cpp
llama-server -hf unsloth/GLM-4.7-Flash-GGUF
vLLM
vllm serve zai-org/GLM-4.7-Flash
MLX (Mac)
mlx_lm.chat --model mlx-community/GLM-4.7-Flash-4bit

Questions

How much VRAM does GLM-4.7 Flash need?

GLM-4.7 Flash needs 17.6 GB for its weights at 4-bit, 18.8 GB with an 8K context and 20.1 GB with 32K. At 8-bit, 32.1 GB.

What is the minimum VRAM for GLM-4.7 Flash?

About 15.4 GB at 3-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 Radeon RX 7900 XTX.

How big is GLM-4.7 Flash?

31.2B parameters (3B active per token): 58.2 GB of weights at 16-bit, 17.6 GB at 4-bit.

Can GLM-4.7 Flash run on an RTX 4090?

Yes. At 5-bit it holds 24K of context and writes about 171 tokens a second.

More models