zai-org/GLM-5.1 VRAM requirements

754B params, Sparse multi-head latent attention

GLM-5.1 needs about 746 GB of GPU memory for FP8 weights. The smallest fitting setup at 32k context is 8× NVIDIA B200. It supports 45 concurrent conversations with SGLang defaults.

GPU requirements by context length

8k context
GPUGPUsConcurrent users
NVIDIA B200864
NVIDIA H200 SXM864
NVIDIA H100 SXM16 (verify server configuration)64
NVIDIA H100 NVL16 (verify server configuration)64
NVIDIA A100 80GB SXM16 (verify server configuration)64
NVIDIA RTX PRO 6000 Blackwell Server Edition16 (verify server configuration)64
NVIDIA H2016 (verify server configuration)64
NVIDIA L40Sno fit
32k context
GPUGPUsConcurrent users
NVIDIA B200845
NVIDIA H200 SXM816
NVIDIA H100 SXM16 (verify server configuration)31
NVIDIA H100 NVL16 (verify server configuration)48
NVIDIA A100 80GB SXM16 (verify server configuration)25
NVIDIA RTX PRO 6000 Blackwell Server Edition16 (verify server configuration)51
NVIDIA H2016 (verify server configuration)51
NVIDIA L40Sno fit
128k context
GPUGPUsConcurrent users
NVIDIA B200811
NVIDIA H200 SXM84
NVIDIA H100 SXM16 (verify server configuration)7
NVIDIA H100 NVL16 (verify server configuration)12
NVIDIA A100 80GB SXM16 (verify server configuration)6
NVIDIA RTX PRO 6000 Blackwell Server Edition16 (verify server configuration)12
NVIDIA H2016 (verify server configuration)12
NVIDIA L40Sno fit

FP8 weights, SGLang defaults, estimates.

Weights by precision

Weight memory
PrecisionWeightsSmallest fitting setup at 32k
BF16about 1488 GB16x NVIDIA B200
FP8about 746 GB8x NVIDIA B200
INT4 (AWQ)about 410 GB4x NVIDIA B200

Model notes

Attention
Sparse multi-head latent attention
Architecture
Mixture-of-Experts
Layers
78
Hidden size
6,144
Vocabulary
154,880

MLA keeps one compact shared KV copy per GPU, so adding GPUs does not shrink the cache, but the copy itself is small for a model this size.

Frequently asked questions

Will zai-org/GLM-5.1 run on a single H100?
No, a single H100 SXM cannot hold zai-org/GLM-5.1 at 32k context with FP8 weights. The model requires multiple GPUs.
What is the cheapest GPU setup for zai-org/GLM-5.1?
The cheapest fitting setup at 32k context is 8x NVIDIA B200 with FP8 weights and SGLang defaults.
How much VRAM does zai-org/GLM-5.1 need at 128k context?
At 128k context on the cheapest fitting setup, zai-org/GLM-5.1 uses about 93 GB of VRAM per GPU for FP8 weights and about 5.9 GB per conversation for the KV cache. The total VRAM needed depends on the GPU count and parallelism configuration.
Can I run zai-org/GLM-5.1 with INT4 quantization?
Yes, zai-org/GLM-5.1 has INT4 weight figures. Its weights take about 410 GB in INT4 versus about 746 GB in FP8. INT4 can make the model fit on fewer GPUs, but check inference quality for your workload.

Want a different context length, GPU, or quantization? The calculator runs the same numbers live, preloaded with this model.

Open in calculator

Don't want to run this yourself? We deploy and operate it for you. Book a call →

Related models

zai-org/GLM-5.2 openai/gpt-oss-120b moonshotai/Kimi-K2.6