How much VRAM does GLM-5.3 need?

About 479 GB at Q4_K_M with a 32K-token context (461 GB weights, 2.9 GB KV cache, 14.8 GB overhead). These are estimates.

Can I run GLM-5.3 on a single GPU?

Not at Q4_K_M with 32K context on any single GPU in our catalog. Use several GPUs, a high-memory unified-memory machine, a smaller quantization or a shorter context.

Which apps can run GLM-5.3 locally?

Verified so far: colibri, llama.cpp, MLX (mlx-lm), KTransformers, vLLM, SGLang, TokenSpeed. Each entry links to its source on this page.

Models / GLM-5.3

Can I run GLM-5.3 locally?

Z.ai · 753B parameters · mixture of experts · up to 1024K context · license other · Hugging Face

At Q4_K_M with a 32K context, GLM-5.3 needs about 479 GB of GPU or unified memory by our estimate. The official checkpoint (FP8) is 756 GB on disk.

Memory by quantization and context

Quantization	Weights	Total at 8K	Total at 32K	Total at 128K
Q3_K_M (~4.0 bit)	376 GB	389 GB	392 GB	400 GB
Q4_K_M (~4.9 bit, popular default)	461 GB	476 GB	479 GB	487 GB
Q5_K_M (~5.7 bit)	537 GB	555 GB	557 GB	566 GB
Q6_K (~6.6 bit)	618 GB	638 GB	641 GB	649 GB
Q8_0 (~8.5 bit, near-lossless)	800 GB	826 GB	828 GB	837 GB
F16 / BF16 (16 bit)	1507 GB	1554 GB	1556 GB	1565 GB

Bits per weight from llama.cpp's quantize docs. F16 KV cache. Totals include our runtime-overhead estimate.

Cheapest hardware that fits (Q4_K_M, 32K)

NVIDIA DGX Station (GB300): about €83,738 (new, configurator.exxactcorp.com, 740 GB usable)

Runs in

Tool	Platforms	How	Source
colibri Disk-streaming MoE engine	Linux, Windows, macOS	coli setup --model glm-5.3 (experts stream from SSD; 16/24 GB RAM)	source
llama.cpp Inference engine (C/C++)	Linux, Windows, macOS, Android (build)	GGUF: unsloth/GLM-5.3-GGUF	source
MLX (mlx-lm) Apple framework + CLI	macOS (Apple Silicon only)	MLX conversion: mlx-community/GLM-5.3-4bit	source
KTransformers CPU+GPU hybrid engine	Linux, Windows	named in the official model card	source
vLLM Serving engine	Linux	named in the official model card	source
SGLang Serving engine	Linux	named in the official model card	source
TokenSpeed Serving engine	Linux	named in the official model card	source

Ollama: cloud only, it does not download the model to your machine.

Only tools with a public source are listed: an official build (ggml-org, lmstudio-community, mlx-community, bartowski or Unsloth on Hugging Face), the model card, ollama.com or the tool's own README. Apps built on llama.cpp, such as Jan and PocketPal AI, can usually load the same GGUF files. All local apps.

Try other settings

ModelQuantizationContext lengthKV cacheMax GPUs

Estimated memory needed479 GB

Weights 461 GBKV cache at 32K 2.9 GBRuntime overhead 14.8 GB

1 hardware options fit

Cheapest real offer per device from AI Hardware Deals, converted to EUR at approximate rates. Multi-GPU rows multiply the card price.

Also show hardware that doesn't fit Include legacy server cards (M40, P40, P100, MI50, MI60)

	Setup		Headroom
NVIDIA DGX Station (GB300) unified memory	1 device	740 GB	261 GB	€83,738 new, configurator.exxactcorp.com	n/a

Speed ceiling is memory bandwidth divided by the bytes of (active) weights read per token, shown only where we have a sourced bandwidth figure. Real decode speed is lower, especially with long contexts or when splitting across cards. All memory numbers are estimates; see the methodology.

FAQ

How much VRAM does GLM-5.3 need?: About 479 GB at Q4_K_M with a 32K-token context (461 GB weights, 2.9 GB KV cache, 14.8 GB overhead). These are estimates.
Can I run GLM-5.3 on a single GPU?: Not at Q4_K_M with 32K context on any single GPU in our catalog. Use several GPUs, a high-memory unified-memory machine, a smaller quantization or a shorter context.
Which apps can run GLM-5.3 locally?: Verified so far: colibri, llama.cpp, MLX (mlx-lm), KTransformers, vLLM, SGLang, TokenSpeed. Each entry links to its source on this page.