EDGE OSZ.AI · 744B PARAMETERS
GLM-5.2 hardware and VRAM requirements
Z.ai’s 744B open-weight coding model with 40B active parameters and a 1M-token context. GLM-5.2 is a mixture-of-experts model with 40B of its 744B parameters work on each token. At FP8 it needs about 783 GB of GPU memory for one user, and about 795 GB to serve a team of 16 with 8K context each.
Total parameters
744B
Active per token
40B
Context length
1M tokens
Released as
FP8, NVFP4 (4-bit)
Minimum VRAM
455 GB at NVFP4
For a team of 16
795 GB at FP8
VRAM needed to run GLM-5.2
GPU memory by precision, number of concurrent users and context length. Each user needs their own KV cache, so memory grows with users and context.
| Precision | Weights only | 1 user, 8K | 16 users, 8K | 16 users, 32K | LoRA fine-tune |
|---|---|---|---|---|---|
| FP8 Released | 744 GB | 783 GB | 795 GB | 835 GB | 904 GB |
| NVFP4 (4-bit) 433 GB official NVFP4 build | 432 GB | 455 GB | 467 GB | 507 GB | 576 GB |
| BF16 Upcast | 1.49 TB | 1.56 TB | 1.58 TB | 1.62 TB | 1.69 TB |
Full fine-tuning in mixed precision needs about 11.98 TB: weights, gradients and optimizer states at roughly 16 bytes per parameter. LoRA trains small adapter layers on a frozen copy of the model instead, so it needs far less.
GPUs that can run GLM-5.2 on-premise
For a team of 16 using GLM-5.2 at FP8 with 8K context each (about 795 GB), smallest fit first.
NVIDIA DGX B200
1.44 TB per system · 501 GB to spare.
AMD Instinct MI325X
256 GB per GPU · Split across 4 GPUs in one server, 126 GB to spare.
NVIDIA H200 NVL
141 GB per GPU · Split across 8 GPUs in one server, 220 GB to spare.
AMD Instinct MI300X
192 GB per GPU · Split across 8 GPUs in one server, 587 GB to spare.
NVIDIA DGX Spark
128 GB per system · Needs more than 2 linked systems.
NVIDIA L40S
48 GB per GPU · Needs more than one 8-GPU server.
NVIDIA H100 NVL
94 GB per GPU · Needs more than one 8-GPU server.
NVIDIA RTX PRO 6000 Blackwell
96 GB per GPU · Needs more than one 8-GPU server.
Some of these setups need several GPUs working as one. Buying the GPUs is the easy part. Vanik Edge OS splits the model across them, keeps it running if a server fails and manages it all from one console. See how Edge OS clusters GPU servers
Weighing two of these? Compare AI hardware side by side, for example H100 vs H200, RTX PRO 6000 vs RTX 5090 or DGX Spark vs Mac Studio.
Running GLM-5.2 for your team?
We will size the GPU server for your users and data, and set up GLM-5.2 on it inside your own building. Free, no obligation.
GLM-5.2 at a glance
Compare GLM-5.2 with other models
FROM SIZING TO RUNNING
Sizing the hardware is the easy part.
Getting several GPUs to work as one system for a whole company is where most projects slow down. That is the part Vanik Edge OS does for you.
STEP 01 · THIS TOOL
Size it
Work out the GPU memory your models need and which GPUs or servers can hold them.
STEP 02 · VANIK EDGE OS
Cluster it
Edge OS joins your GPU servers into one cluster, splits a large model across them and keeps it running if a server fails. NVIDIA and AMD servers work side by side.
STEP 03 · VANIK EDGE OS
Run it
Your team gets AI apps with single sign-on and an audit trail, managed from one console, with your data staying inside your building.
GLM-5.2 GPU questions
Can GLM 5.2 see images?
No. GLM-5.2 takes text input and produces text output, according to its NVIDIA listing. For image and video input, Z.ai’s GLM-5.3-Flash is multimodal.
What are the GLM 5.2 hardware requirements?
To run GLM-5.2 locally you need about 455 GB of GPU memory at NVFP4 for one user (795 GB for a team of 16), system RAM of at least 432 GB so the weights can be loaded, around 432 GB of fast NVMe storage for the model files, and an inference engine such as vLLM or SGLang. A team of 16 fits on NVIDIA DGX B200.
Is GLM 5.2 open source?
Yes. Z.ai released GLM-5.2’s weights openly, in FP8 and NVFP4 formats, under the MIT license, so you can download and self-host it.
How much VRAM to run GLM 5.2?
GLM-5.2 needs about 783 GB of GPU memory at FP8 for one user with 8K context: 744 GB for the weights plus the KV cache and runtime overhead. Serving 16 people at once with 8K context each needs about 795 GB.
Can you run GLM 5.2 locally?
Yes, on data-center GPUs. At NVFP4 it needs about 455 GB of GPU memory for one user, more than consumer graphics cards or laptops hold. A team of 16 fits on NVIDIA DGX B200.
How big is GLM 5.2?
GLM-5.2 has 744B parameters in total, of which 40B work on each token. The weights take about 432 GB at NVFP4.
Why does GLM 5.2 need so much memory if only 40B parameters are active?
It is a mixture-of-experts model. Only 40B parameters work on each token, which keeps it fast, but a different set of experts is picked for every token, so all 744B parameters must stay loaded in GPU memory.
Can I fine-tune GLM 5.2?
LoRA fine-tuning at FP8 needs roughly 904 GB with 4K sequences; it trains small adapter layers on a frozen copy of the model. Full fine-tuning needs about 11.98 TB, which calls for a multi-GPU training cluster.
How are these numbers calculated?
Weights are parameters × bytes per parameter. The KV cache uses 97.7 KB per token per user. We add about 5% of the weights plus 1 GB for runtime overhead and keep 10% of each GPU free. Figures are planning estimates; real usage depends on the inference engine and settings.
VRAM for similar models
Specifications from Hugging Face: zai-org/GLM-5.2 and NVIDIA NGC: GLM-5.2. Memory and bandwidth are manufacturer figures; GPU counts are rounded to 1, 2, 4 or 8 per server. All figures are planning estimates. Last checked November 10, 2026.
Run GLM-5.2 inside your own building
Vanik Edge OS runs open-weight models on NVIDIA DGX, AMD Instinct and other GPU servers on your premises, so your data never leaves.