RunsOnMy ModelsGPUs & MacsBy VRAMAPI pricesVRAM calculatorNew

Can GLM 4.6V run on the H200?

Yes. At 8K context the best fit is Q8_0, using about 108 GB of the 134 GB usable on the H200. Estimated speed ceiling: ~319 tokens/sec.

Memory by quant and context

Quant4K ctx8K ctx32K ctx128K ctxMax tok/s (est., 8K)
2-bit (Q2_K) •42.3 GB Fits43.0 GB Fits47.3 GB Fits64.5 GB Fits825
3-bit (Q3_K_M) •53.0 GB Fits53.8 GB Fits58.1 GB Fits75.3 GB Fits653
4-bit (Q4_K_M) •67.1 GB Fits67.8 GB Fits72.2 GB Fits89.4 GB Fits513
5-bit (Q5_K_M) •76.6 GB Fits77.3 GB Fits81.6 GB Fits98.9 GB Fits448
6-bit (Q6_K) •90.7 GB Fits91.4 GB Fits95.7 GB Fits113 GB Fits378
8-bit (Q8_0) •107 GB Fits108 GB Fits112 GB Fits129 GB Tight319
16-bit (F16/BF16) •200 GB No201 GB No206 GB No223 GB No—

• real GGUF file size from unsloth/GLM-4.6V-GGUF. Speed is a bandwidth ceiling (4800 GB/s), not a benchmark. Methodology.

More

GLM 4.6V 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-4.6V), NVIDIA specs. Data as of 2026-11-17.