Can I Run Nemotron-Cascade 2 30B-A3B on NVIDIA H100 80GB (PCIe)?

Written by Jakub Rusinowski · Last updated September 11, 2026

Yes. Nemotron-Cascade 2 30B-A3B needs 21.7 GB at 8K context (19.1 GB weights + 2.6 GB KV cache/overhead) at Q4_K_M and the NVIDIA H100 80GB (PCIe) has 80 GB, leaving about 58.3 GB of headroom, at roughly 249 tok/s (estimated) with room for up to 256K of context.

See what else this hardware can run →

The numbers

Nemotron-Cascade 2 30B-A3B on the NVIDIA H100 80GB (PCIe): VRAM by quantization

Assumes an 8K-token context with an f16 KV cache, on an estimated architecture — this model publishes no config we can read. A longer window needs more; a quantized KV cache needs less. “Max context” is the largest window that still fits in 80 GB. Figures are estimates from parameter count, quantization and memory bandwidth — the analyzer lets you tune KV-cache quant and context.

Context costs VRAM too. At Q4_K_M on the NVIDIA H100 80GB (PCIe): 8K 21.7 GB · 32K 27.1 GB · 128K 48.9 GB. Every window this model publishes still fits 80 GB.

What to do next

Nothing to change — run it at Q4_K_M. If you want more capability, check whether the next size up also fits.

No compatible GPU? Nemotron-Cascade 2 30B-A3B on 24 GB of system RAM, CPU only: runs.

Running Nemotron-Cascade 2 30B-A3B on this card:

Nemotron-Cascade 2 30B-A3B on nearby hardware

• Apple M5 Pro (64 GB) — runs
• Apple M2 Max (96 GB) — runs

All Nemotron Cascade 2 sizes on the NVIDIA H100 80GB (PCIe) | Nemotron-Cascade 2 30B-A3B model page | NVIDIA H100 80GB (PCIe) GPU page | VRAM calculator | Check your hardware