RunsOnMy ModelsGPUs & MacsBy VRAMBest GPUCan I run…?API pricesVRAM calculatorNew

Can GLM 4.6 run on the H100 SXM?

Not fully in VRAM. Even at 2-bit with 8K context, GLM 4.6 needs about 125 GB, more than the 76.0 GB usable on the H100 SXM. It can still run with offload (slower): Q3_K_M with part of the model in system RAM, about 4.8–9.7 tokens/sec (estimate). Details and command.

Runs with offload (slower)

Runs with offload 3-bit (Q3_K_M) at 8K context: all attention layers plus the experts of 38 of 90 MoE layers on the H100 SXM (about 74.3 GB), and the experts of the other 52 layers in system RAM (about 87.4 GB). Needs a PC with at least 128 GB of system RAM. Estimated speed: 4.8–9.7 tokens/sec (slow but usable), versus a bandwidth ceiling of ~206 tok/s if it fit fully in VRAM.

llama-server -hf unsloth/GLM-4.6-GGUF:Q3_K_M -c 8192 -ngl 999 --n-cpu-moe 54

Estimate, not a benchmark. The command downloads the GGUF from Hugging Face. The speed range spans dual-channel DDR4-3200 to DDR5-6000 system RAM; your CPU, RAM speed and llama.cpp version move it. If llama.cpp runs out of GPU memory, raise --n-cpu-moe by 1-2; if VRAM is left over, lower it for more speed. Recent llama.cpp builds can also pick the split automatically. How we estimate offload.

Memory by quant and context

Quant4K ctx8K ctx32K ctx128K ctxMax tok/s (est., 8K)
2-bit (Q2_K) •124 GB No125 GB No134 GB No168 GB No—
3-bit (Q3_K_M) •161 GB No162 GB No171 GB No206 GB No—
4-bit (Q4_K_M) •203 GB No204 GB No213 GB No247 GB No—
5-bit (Q5_K_M) •238 GB No239 GB No248 GB No282 GB No—
6-bit (Q6_K) •275 GB No277 GB No285 GB No320 GB No—
8-bit (Q8_0) •355 GB No357 GB No365 GB No400 GB No—
16-bit (F16/BF16) •667 GB No668 GB No677 GB No711 GB No—

• real GGUF file size from unsloth/GLM-4.6-GGUF. Speed is a bandwidth ceiling (3350 GB/s), not a benchmark. Methodology.

Hardware that does fit it

H200Mac M2 Ultra 192GBMac M3 Ultra 256GBMac M3 Ultra 512GB

What fits instead

"Runs" and "fits" mean the whole model in GPU memory at 8K context (4-bit or better for the model pick). Prices are US launch MSRPs, not today's street prices. "Popular" means the 60 most-downloaded models we track.

More

GLM 4.6 on every GPU and Mac · Every model that runs on the H100 SXM · Other models that fit 64 GB VRAM · Try other settings in the calculator

Guides: Best GPU for local LLMs by budget · Can I run gpt-oss, Llama 70B or Qwen3? · Mac vs PC for local LLMs

Data: Hugging Face Hub (zai-org/GLM-4.6), NVIDIA specs. Model data last checked 2026-11-17; refreshed hourly.