AMD MI325X GPU: Specs, VRAM, Price & Benchmarks (2026)
The AMD MI325X is a real GPU (launched 2024), but no provider on Aquanode is listing it for rental right now, so there is no live hourly price to quote. Availability changes as providers add and retire hardware; the specs, VRAM and price context below still apply.
How much VRAM does the AMD MI325X have?
The AMD MI325X has 256GB HBM3E, with 6 TB/s of peak memory bandwidth.
AMD MI325X VRAM calculator: check which models fit in its memory at each precision.
All models that fit in 192 GB: the open models whose weights and overhead fit, at native, FP8 and INT4 precision.
What can the AMD MI325X run?
Popular open models from small to frontier scale, with the memory each needs and how many AMD MI325X cards (256GB HBM3E each) that takes.
| Model | As published | FP8 | INT4 |
|---|---|---|---|
| Qwen/Qwen3-8B 8.2B | BF16: ~18.3 GB, 1 GPU | FP8: ~9.2 GB, 1 GPU | INT4: ~4.6 GB, 1 GPU |
| Qwen/Qwen2.5-14B-Instruct 14.8B | BF16: ~33 GB, 1 GPU | FP8: ~16.5 GB, 1 GPU | INT4: ~8.3 GB, 1 GPU |
| Qwen/Qwen3-32B 32.8B | BF16: ~73.2 GB, 1 GPU | FP8: ~36.6 GB, 1 GPU | INT4: ~18.3 GB, 1 GPU |
| Qwen/Qwen-72B 72.3B | BF16: ~162 GB, 1 GPU | FP8: ~80.8 GB, 1 GPU | INT4: ~40.4 GB, 1 GPU |
| MiniMaxAI/MiniMax-M2.7 228.7B | FP8: ~256 GB, 1 GPU | – | INT4: ~128 GB, 1 GPU |
| deepseek-ai/DeepSeek-R1 684.5B | FP8: ~765 GB, 3 GPUs | – | INT4: ~383 GB, 2 GPUs |
Estimates: weights at the stated precision plus a flat 20% for KV cache and overhead, at a moderate context length. A dash means the precision is not offered for that model (it is already published at that size). INT4 needs a published quantized checkpoint. Open any model for a per-GPU breakdown, or use the AMD MI325X VRAM calculator.
AMD MI325X specs
| Architecture | launched 2024 |
| VRAM | 256GB HBM3E |
| Memory bandwidth | 6 TB/s |
| FP16 / BF16 tensor throughput | 1,307.4 TFLOPS (peak, dense) |
| FP8 tensor throughput | 2,614.9 TFLOPS (peak, dense) |
| Interconnect | Infinity Fabric, 8 links at 128 GB/s peak each |
| TDP | 1000W peak |
| Form factor | OAM |
Specs sourced from the vendor's public product page. See the source.
GPU Glossary: What is VRAM?, HBM, Tensor Cores, CUDA Cores, TFLOPS, NVLink vs PCIe
Good for
256GB on a single card (more than three times an H100's 80GB), so a 70B model fits in FP16 with room to spare and much larger models need far fewer GPUs, at a memory bandwidth (6 TB/s) ahead of H200's 4.8 TB/s. Memory-bound inference of big models is where the capacity pays off.
Not good for
Runs on AMD's ROCm software stack rather than CUDA, so some CUDA-only tooling and custom kernels need a ROCm port or do not run at all. Check your stack's ROCm support first.
Related guides
Other models in the same generation, or see how it compares in the GPU benchmarks and specs table. The full list is in the GPU index.
Related reading: AMD MI325X guide.
Get notified when the price drops
GPU supply moves hourly. Tell us what you're waiting for and we'll email you when a matching offer appears across any provider we track.
