Models / GLM-5.3

Best home rig for GLM-5.3

Current model753B dense1 measured result model info ↗

Will it fit my GPU? Best value GPU per euro →

Real, measured generation speeds for GLM-5.3 on home hardware, fastest first. Different quants/engines aren't directly comparable.

Build	Quant	Engine	tok/s	Prompt processing (tok/s)	Cost	Source
3x DGX Spark: GLM-5.3 EXL3 2.75-bit on TensorFold, 31.3 tok/s (Mia's AI Lab recipe)	EXL3 2.75-bit + DSpark	TensorFold, 3 nodes, 1 stream	31.3 ⚡	591 @32k · est. TTFT @8k ≈13.9 s*		mia-ai.net ↗

Prompt processing = prefill speed reported by the source, @ prompt length. * est. TTFT @8k is calculated from measured prefill, not measured. Why it matters.

Rent before you buy: try this model on a cloud GPU for a few dollars first: Runpod (referral link; sign up with Google to get the new-user bonus)

Memory math: how much VRAM / RAM does GLM-5.3 need?

GLM-5.3 has 753B parameters (source: Hugging Face listing: 753B). Weight size = parameters × bits-per-weight ÷ 8, before context (KV cache) and runtime overhead.

Format	Bits/weight	Weights	Smallest common memory size that holds the weights alone
Q8_0	8.50	800.1 GB	1024 GB
Q6_K	6.56	617.8 GB	1024 GB
Q5_K_M	5.70	536.9 GB	1024 GB
Q4_K_M	4.89	460.7 GB	512 GB
Q3_K_M	4.00	376.1 GB	512 GB
Q2_K	3.16	297.4 GB	512 GB

Bits-per-weight for GGUF quants are llama.cpp's own figures (llama.cpp quantize README). "As measured" rows apply the source's stated bits-per-weight to the parameter count; real files differ slightly because some layers stay at higher precision. Try other combinations in the VRAM calculator.

Measured on setups with less GPU memory than the Q4_K_M weights (so part of the model lived in system RAM or was streamed from disk):

3x Nvidia GB10 (unified, DGX Spark) (384 GB) + 128 GB LPDDR5X unified per Spark (384 GB total): 31.3 tok/s (EXL3 2.75-bit + DSpark)

FAQ

How much VRAM does GLM-5.3 need?

The weights are about 460.7 GB at Q4_K_M (753B params × 4.89 bits ÷ 8, before context/KV cache). Add memory for context (KV cache) and the runtime.

What is the fastest measured hardware for GLM-5.3?

3x DGX Spark (GB10): 31.3 tok/s generation (TensorFold, 3 nodes, 1 stream, EXL3 2.75-bit + DSpark), from https://mia-ai.net/models/GLM-5.3-EXL3-3x-DGX-Sparks-TensorFold.

Which GPUs have measured results for GLM-5.3?

3x DGX Spark (GB10).

Can it run on… 3x DGX Spark (GB10)