NVIDIA · 141 GB HBM3e · 4800 GB/s bandwidth ·
Google · 11.95B · model ctx up to 256K
Yes — the H200 141GB SXM5 (141 GB HBM3e) can run gemma-4-12b-it. At Q4 with the default 8K context it needs ≈ 9.96 GB VRAM (weights 6.57 GB + KV cache 2.39 GB + 1 GB runtime overhead), which fits within 141 GB. Decode at Q4/8K: ≈ 255.6 tok/s (estimated, batch 1). FP16 (≈ 27.29 GB) also fits, at ≈ 70.3 tok/s (estimated). Estimates come from memory-bandwidth math; rows tagged measured override estimates. Relative ranking is more reliable than absolute tok/s. Methodology.
Fit & speed by quant and context
Quant
Context
VRAM needed
Fits?
tok/s (decode)
Q4
4K
8.77 GB
Yes
255.6est.
Q4
8K default
9.96 GB
Yes
255.6est.
Q4
32K
17.13 GB
Yes
255.6est.
FP16
4K
26.09 GB
Yes
70.3est.
FP16
8K default
27.29 GB
Yes
70.3est.
FP16
32K
34.46 GB
Yes
70.3est.
Fits = weights + KV(ctx) + 1 GB overhead ≤ 95% of VRAM. Measured
anchors are context-agnostic; the fit verdict is recomputed per context. A missing tok/s
means the model is far beyond this card (offload-only territory).