NVIDIA H100 SXM
The H100 SXM is NVIDIA's flagship datacenter GPU for AI training, built on the Hopper architecture with 4th-gen Tensor Cores. With 80GB HBM3 and 900GB/s NVLink, it excels at large-scale distributed training.
VRAM
80 GB
Memory
HBM3
Bandwidth
3,350 GB/s
FP16
1,979 TF
TDP
700 W
Memory and compute
Architecture and power
Notes
Flagship datacenter GPU for AI training
Models that may fit on NVIDIA H100 SXM
Estimates based on INT8 quantization. Actual fit depends on framework and batch size.
More GPUs by NVIDIA
The manufacturer's other cards in the catalog, largest memory first, with bandwidth and power where published.
Frequently asked questions about NVIDIA H100 SXM
Answered from the entry's own figures: memory, which models fit, bandwidth, power and the card's class.
NVIDIA H100 SXM has 80 GB of HBM3 at 3,350 GB/s. At INT8 a model's weights take about one byte per parameter, so the card holds roughly a 64B-parameter model with room for context; larger models need several cards or a lower precision.
From the catalog's estimates at INT8, Llama 3_3 Nemotron Super 49B V1, Gemma 4 31B IT, Llama Nemotron Embed VL 1B V2 and GLM Z1 9B 0414 fit on a single NVIDIA H100 SXM with their default context. The list on this page shows the estimated memory each takes; the capacity planner sizes any model against this card for your own context length, batch size and request rate.
3,350 GB/s. Bandwidth bounds how fast a model generates tokens, because every token reads the whole set of weights and the KV cache from memory; at this rate a 64B model at INT8 could be read on the order of 52 times a second on one card, before batching and framework efficiency.
NVIDIA H100 SXM has a TDP of 700 W, with a maximum of 700 W, in the SXM5 form factor. Budget the TDP per card plus host overhead when sizing a server's power and cooling.
NVIDIA H100 SXM is a datacenter accelerator, meant for servers and multi-GPU serving, built on the Hopper architecture and released in 2022. It supports a multi-GPU interconnect at 900 GB/s, so a model larger than one card can be split across several.
Specifications from the manufacturer's published material; a figure that is not published is shown as such, and a derived one is explained in the notes.
Added Jan 25, 2026 · Last updated Jan 25, 2026
Explore more
Companion tools that draw on the same catalog and routing engine.
Put intelligence in the middle.
One gateway in front of every model, with your policies applied and every decision on record. Start on the free plan with 5,000 requests a month, no card required.