RunsOnMy ModelsGPUs & MacsBy VRAMAPI pricesVRAM calculatorNew

Can GLM 4.5 run on the H200?

Yes. At 8K context the best fit is Q2_K, using about 126 GB of the 134 GB usable on the H200. Estimated speed ceiling: ~369 tokens/sec.

Memory by quant and context

Quant4K ctx8K ctx32K ctx128K ctxMax tok/s (est., 8K)
2-bit (Q2_K) •124 GB Tight126 GB Tight134 GB No169 GB No369
3-bit (Q3_K_M) •162 GB No163 GB No172 GB No206 GB No—
4-bit (Q4_K_M) •204 GB No205 GB No214 GB No248 GB No—
5-bit (Q5_K_M) •239 GB No240 GB No249 GB No283 GB No—
6-bit (Q6_K) •276 GB No278 GB No286 GB No321 GB No—
8-bit (Q8_0) •357 GB No358 GB No367 GB No401 GB No—
16-bit (F16/BF16) •670 GB No671 GB No680 GB No714 GB No—

• real GGUF file size from unsloth/GLM-4.5-GGUF. Speed is a bandwidth ceiling (4800 GB/s), not a benchmark. Methodology.

More

GLM 4.5 on every GPU and Mac · Every model that runs on the H200 · Other models that fit 64 GB VRAM · Try other settings in the calculator

Data: Hugging Face Hub (zai-org/GLM-4.5), NVIDIA specs. Data as of 2026-11-17.