Two AMD data center GPUs, one CDNA 3 and one CDNA 4: 256 GB HBM3e against 288 GB HBM3e and 33% more memory bandwidth on the MI350X.
At a glance
Memory
- MI325X
- 256 GB
- MI350X
- 288 GB
1.1x more on MI350X
Memory bandwidth
- MI325X
- 6.00 TB/s
- MI350X
- 8.00 TB/s
1.3x higher on MI350X
Cheapest on-demand
- MI325X
- $2.25 /GPU/hr
- MI350X
- --
MI350X is not sold on demand
Price comparison
Median price per GPU per hour by billing type, with the cheapest listing under it.
| Billing | MI325X | MI350X | Difference |
|---|---|---|---|
| On-demand | $3.06 /GPU/hr from $2.25 | No on-demand listings | |
| Reserved | $1.95 /GPU/hr from $1.48 (12 mo) | -- /GPU/hr from $4.76 (12 mo) | |
| Spot | Sold out | -- /GPU/hr from $4.00 |
Heads up: A provider's own page may quote a different figure, for example on tax, region or a promotion. A monthly-only plan shows a derived hourly rate. Medians are across providers, one vote each, so a large catalog does not outweigh a small one. No median for reserved and spot on the MI350X: we only show one when at least 3 providers list the GPU on that billing type. Verify before provisioning. More on how we price.
Price history
MI325X vs MI350X price history
Weekly median on-demand price per GPU per hour
Specs comparison
| Spec | MI325X | MI350X | Difference |
|---|---|---|---|
| Memory | |||
| Capacity | 256 GB HBM3e | 288 GB HBM3e | 1.1x more on MI350X |
| Bandwidth | 6,000 GB/s | 8,000 GB/s | 1.3x higher on MI350X |
| Bus width | 8,192-bit | 8,192-bit | Identical |
| Compute · dense | |||
| FP4 | Not supported | 9,227.5 TFLOPS | MI350X only |
| FP8 | 2,614.9 TFLOPS | 4,614 TFLOPS | 1.8x higher on MI350X |
| FP16 / BF16 | 1,307.4 TFLOPS | 2,309.6 TFLOPS | 1.8x higher on MI350X |
| INT8 | 2,614.9 TOPS | 4,614 TOPS | 1.8x higher on MI350X |
| FP32 | 163.4 TFLOPS | 144.2 TFLOPS | 1.1x higher on MI325X |
| FP64 | 81.7 TFLOPS | 72.1 TFLOPS | 1.1x higher on MI325X |
| Platform | |||
| Interconnect | Infinity Fabric 896 GB/s | Infinity Fabric 1,075 GB/s | 1.2x higher on MI350X |
| Board power | 1,000 W | 1,000 W | Identical |
| Released | Q4 2024 | Q2 2025 | 8 months newer on MI350X |
| Precision | |||
| Supported formats | FP8FP16BF16TF32FP32FP64INT8 | FP4FP6FP8FP16BF16FP32FP64INT8 | |
Compute: Vendor peak figures, dense and per GPU. Not measured throughput.
Source: MI325X datasheet and MI350X datasheet.
Where to rent them
Providers listing these GPUs, those with both first, so you can switch without moving clouds.
| Provider | MI325X /GPU/hr | MI350X /GPU/hr |
|---|---|---|
DigitalOcean
|
On-Demand $3.80 | Reserved (12mo) $4.76 |
Bentaus
|
On-Demand $2.25 | -- |
Cyfuture AI
|
On-Demand $3.01 | -- |
TensorWave
|
Custom $2.25 | -- |
Vultr
|
Spot $2.00 | -- |
Similar GPUs
Other comparisons in the same class.
-
Nvidia H100
80 GB HBM3 · 44% less bandwidth than the MI325X
From $1.25 /GPU/hr Compare -
Nvidia B200
180 GB HBM3e · 28% more bandwidth than the MI325X
From $3.75 /GPU/hr Compare
-
Nvidia H200
141 GB HBM3e · 20% less bandwidth than the MI325X
From $2.09 /GPU/hr Compare
Our data for AMD MI325X was last updated on Sept. 14, 2026, and for AMD MI350X on Sept. 14, 2026.