Side-by-side GPU comparison: specs, memory, compute performance, and live cloud pricing.
Verdict
MI325X has 112 GB more VRAM, making it better suited for large models and long context windows. For compute-bound workloads like training, MI325X delivers 48.1× higher FP16 throughput. MI325X supports a broader range of models (671 vs 657 from this catalog), giving more flexibility.
Specifications
| M3 Ultra (192 GB) | MI325X | |
|---|---|---|
| VRAM | 144 GB | 256 GB |
| VRAM Type | LPDDR5 | HBM3e |
| Memory Bandwidth | 0.8 TB/s | 6.0 TB/s |
| FP16 Performance | 27 TFLOPS | 1307 TFLOPS |
| Manufacturer | Apple | AMD |
| FP8 Support | No | Yes |
| FP4 Support | No | No |
Price / Performance
Based on cheapest single-GPU on-demand pricing. Lower $/TFLOP = better compute value; lower $/GB = better memory value.
| M3 Ultra (192 GB) | MI325X | |
|---|---|---|
| $/hr (cheapest) | — | — |
| $/TFLOP (compute value) | — | — |
| $/GB VRAM (memory value) | — | — |
Cloud Pricing
Cheapest on-demand price per provider (single GPU).
M3 Ultra (192 GB)
No cloud pricing available.
MI325X
No cloud pricing available.
Model Compatibility
Models from the catalog that fit on each GPU, grouped by required precision.
M3 Ultra (192 GB) (657 models)
MI325X (671 models)
You might also compare…
Pricing data refreshed hourly · Last updated April 11, 2026 · Browse all comparisons