H100 80GB SXM5

NVIDIA · 80 GB HBM3 · 3350 GB/s · 700 W · MSRP $30,000

The H100 80GB SXM5 is a NVIDIA accelerator with 80 GB HBM3 and about 3350 GB/s peak memory bandwidth. AI Hashrate estimates local LLM decode (batch 1) at 8K context: fit counts weights + KV cache + 1 GB runtime overhead — not weights alone. At Q4, about 46 curated models fit fully on this card; at FP16, about 37 fit. Top Q4 speeds: Qwen3-0.6B ≈ 2664.8 tok/s (estimated, fits); gemma-3-1b-it ≈ 2131.8 tok/s (estimated, fits); Llama-3.2-1B-Instruct ≈ 1776.5 tok/s (estimated, fits); Qwen3-1.7B ≈ 1065.9 tok/s (estimated, fits); Qwen3-30B-A3B ≈ 710.6 tok/s (estimated, fits). Relative ranking is more reliable than absolute tok/s. See methodology for the bandwidth formula and measured-anchor policy. Methodology.

Table: decode speed estimates (Q4 / FP16). Measured rows override estimates. MSRP is list price, not live retail. Context default 8K.