COMPARE / H200 SXM VS MI325X

H200 SXM vs MI325X

Specs from the chip makers' datasheets; prices from the live Compute Cafe panel, refreshed hourly. Every rate is an on-demand published price with a source receipt.

CHEAPEST H200 SXM
$0.774/hr
OVHcloud · median $4.59 · 58 providers
CHEAPEST MI325X
$1.590/hr
SpotGPUs · median $2.00 · 4 providers
CHEAPER RIGHT NOW
H200 SXM
live panel, 2026-09-27
H200 SXMMI325X
MAKERNVIDIAAMD
ARCHITECTUREHopperCDNA 3
LAUNCHED20242024
MEMORY141 GB HBM3e256 GB HBM3e
MEMORY BANDWIDTH4.8 TB/s6 TB/s
TENSOR COMPUTE989 TFLOPS FP16/BF16 tensor dense (1,979 sparse); 1,979 TFLOPS FP8 dense1.3 PFLOPS FP16/BF16 dense
POWER DRAW700 W1,000 W (OAM)
INTERCONNECTNVLink 4.0 900 GB/sInfinity Fabric

HOW TO CHOOSE

MI325X vs H200 SXM
The H200's 141 GB is the CUDA-ecosystem answer to large-model serving. The MI325X nearly doubles the memory; the trade is ROCm versus CUDA.
Workload fit
H200 SXM: large-model inference, memory-bound training, moe serving. MI325X: large-model inference, training, cuda-locked stacks.

H200 SXM VS MI325X - FAQ

Which is cheaper to rent, H200 SXM or MI325X?
By today's published on-demand rates: H200 SXM starts at $0.774/GPU/hr (median $4.59, 58 providers) and MI325X starts at $1.590/GPU/hr (median $2.00, 4 providers). The cheapest live listing is the H200 SXM at OVHcloud. Figures recompute from the panel on every refresh.
H200 SXM or MI325X - which should I rent?
H200 SXM: The H200 is an H100 with the memory subsystem rebuilt: 141 GB of HBM3e at 4.8 TB/s versus 80 GB at 3.35. MI325X: The MI325X is the memory-maximized refresh of the MI300X: same CDNA 3 compute, a third more HBM3e at 256 GB. The side-by-side spec table and live price rows above give the full picture.
Where can I rent H200 SXM and MI325X?
58 providers publish H200 SXM rates and 4 publish MI325X rates on Compute Cafe. Every row links to the provider's own published price - committed-use and reserved rates are excluded.

Full pricing: H200 SXM · MI325X · all comparisons