← GPU price tracker

MI300X GPU Rental Price

The AMD MI300X (CDNA 3, 192 GB HBM3) rents from $1.65/hr, averaged across 4 cloud providers, updated 2026-08-14. On-demand pricing has risen 50% over the past 90 days.

$3.56
on-demand /hr
$1.65
spot /hr
$2.52
reserved /hr
$2,599
/mo on-demand

AMD's H100/H200 competitor — its large 192 GB memory makes it attractive for memory-bound inference of big models. CDNA 3 · 192 GB HBM3 · 5.3 TB/s memory bandwidth · launched 2023.

on-demand /hr · spot /hr

Recent daily prices

DateOn-demandSpot
2026-08-14$3.56$1.65
2026-08-13$3.56$1.65
2026-08-12$3.56$1.65
2026-08-11$3.56$1.65
2026-08-10$3.56$1.65
2026-08-09$3.47$1.65
2026-08-08$3.56$1.65
2026-08-07$3.56$1.65
2026-08-05$3.56$1.65
2026-08-04$3.56$1.65
2026-08-03$3.56$1.65
2026-08-02$3.56$1.65
2026-08-01$3.51$1.65
2026-07-31$3.51$1.65

MI300X vs other GPUs

GPUOn-demand /hrSpot /hrMemory
MI300X$3.56$1.65192 GB HBM3
B200$6.84$3.74192 GB HBM3eMI300X vs B200 →
H200$4.37$2.98141 GB HBM3eMI300X vs H200 →
H100$4.00$1.7180 GB HBM3MI300X vs H100 →
MI355X$8.60$5.29288 GB HBM3eMI300X vs MI355X →
MI325X$3.18$2.00256 GB HBM3eMI300X vs MI325X →

FAQ

How much does it cost to rent an MI300X per hour?

The MI300X averages $3.56/hr on-demand, $1.65/hr spot, and $2.52/hr reserved across 4 cloud providers (updated 2026-08-14).

What is the cheapest MI300X cloud price?

The lowest tier is spot at about $1.65/hr on average; reserved (1–3 year commitment) runs ~$2.52/hr. Compare live per-provider rates on the Bargo GPU price tracker.

How much is an MI300X per month?

At on-demand $3.56/hr, one MI300X runs about $2,599/month (730 hours); reserved pricing lowers that to roughly $1,840/month.

Is the MI300X cheaper than the H100?

On-demand, the MI300X is $3.56/hr vs the H100 at $4.00/hr — cheaper per hour. See the full MI300X vs H100 comparison for specs and cost-per-job context.

What's the difference between MI300X spot and on-demand pricing?

Spot ($1.65/hr) is interruptible spare capacity at a discount to on-demand ($3.56/hr); use spot for checkpointed or fault-tolerant jobs and on-demand or reserved for long, uninterrupted runs.

Cross-provider averages from public cloud GPU listings. Informational only.