AMD MI300X GPU: Specs, VRAM, Price & Benchmarks (2026)
The AMD AMD MI300X (192GB HBM3) launched in 2023. This guide covers its full specs, VRAM, SXM/PCIe differences where they apply, AI performance, and live per-GPU rental pricing on Aquanode.
Short answer on cost: the AMD MI300X rents from $2.63 per GPU per hour on Aquanode. Out of capacity right now.
How much VRAM does the AMD MI300X have?
The AMD MI300X has 192GB HBM3, with 5.3 TB/s of peak memory bandwidth.
- VRAM: AMD MI300X 192GB HBM3 vs H100 80GB HBM3
- VRAM: AMD MI300X 192GB HBM3 vs B200 180GB HBM3e
What fits in 192GB HBM3 of VRAM
| Model | Precision | Fits? |
|---|---|---|
| Llama 3.1 70B | BF16 | ~140GB. Fits on a single card with room to spare. |
| Llama 3.1 405B | FP8 | ~410GB. Needs 3+ cards even at 192GB each. |
| Llama 3.1 405B | INT4 | ~200-230GB, more than its 192GB. Needs 2 cards. |
| Mixtral 8x22B | BF16 | ~280GB. Needs 2 cards. |
Approximate, based on published parameter counts and standard bytes-per-parameter rules of thumb (FP16 ≈ 2 bytes/param, INT4 ≈ 0.5-0.6 bytes/param). Real footprint also depends on KV-cache size and framework overhead.
What can the AMD MI300X run?
Popular open models from small to frontier scale, with the memory each needs and how many AMD MI300X cards (192GB HBM3 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, 2 GPUs | – | INT4: ~128 GB, 1 GPU |
| deepseek-ai/DeepSeek-R1 684.5B | FP8: ~765 GB, 4 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 MI300X VRAM calculator.
AMD MI300X 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.
AMD MI300X specs
| Architecture | AMD, launched 2023 |
| VRAM | 192GB HBM3 |
| Memory bandwidth | 5.3 TB/s |
| FP16 / BF16 tensor throughput | 1,307.4 TFLOPS (peak, dense) |
| FP8 tensor throughput | 2,614.9 TFLOPS (peak, dense) |
| Interconnect | Infinity Fabric |
| TDP | 750W |
| Form factor | OAM |
Specs sourced from the vendor's public datasheet/product page. See the source.
Related reading: ROCm vs CUDA, AMD vs NVIDIA GPUs for AI, AMD MI355X guide, and The best GPUs for AI, ranked.
GPU Glossary: What is VRAM?, HBM, Tensor Cores, CUDA Cores, TFLOPS, NVLink vs PCIe
AMD MI300X AI performance
192GB on a single card (more than double an H100's 80GB), so it fits a 70B model in FP16 with room to spare and holds larger models without splitting across GPUs, at a memory bandwidth (5.3 TB/s) ahead of H100's 3.35 TB/s.
- Dense FP16/BF16 tensor throughput: 1,307.4 TFLOPS
- Dense FP8 tensor throughput: 2,614.9 TFLOPS
- Memory bandwidth: 5.3 TB/s
Runs on AMD's ROCm software stack rather than CUDA. Most inference/training frameworks support it, but some CUDA-only tooling and custom kernels need a ROCm port or don't run at all. Worth checking your stack's ROCm support before committing to it.
See how it stacks up against other cards in the GPU benchmarks and specs table.
AMD MI300X price: what does it cost?
Renting. The cheapest current on-demand rate for the AMD MI300X on Aquanode is $2.63/GPU/hr. Live rates range from $2.63 to $3.29 per GPU per hour, with a median of $2.63/GPU/hr. Billed by the offer's own terms; the table below shows every live rate. Out of capacity right now.
| Region | $/GPU/hr | Available | VRAM | vCPU | RAM |
|---|---|---|---|---|---|
| – | $2.63 | 0 | 192 GB | – | – |
| Michigan, Us | $3.29 | 1 | 192 GB | 8 | 224 GB |
AMD MI300X price history
Aquanode stores one snapshot of its GPU price index per UTC day. For the AMD MI300X that is 5 days so far, 2026-10-06 to 2026-10-10, so this is a short history, not a long-run trend. The lowest per-GPU rate was $2.63 on both 2026-10-06 and 2026-10-10.
| Day (UTC) | Lowest per GPU hour | Median per GPU hour | Data-center lowest per GPU hour | Offers |
|---|---|---|---|---|
| 2026-10-10 | $2.63 | $2.63 | $2.63 | 9 |
| 2026-10-09 | $2.63 | $2.63 | $2.63 | 9 |
| 2026-10-08 | $2.63 | $2.63 | $2.63 | 9 |
| 2026-10-07 | $2.63 | $2.63 | $2.63 | 10 |
| 2026-10-06 | $2.63 | $2.63 | $2.63 | 9 |
Each row is the stored daily snapshot of the live index, copied as recorded. A dash means that day stored no figure. The same series is available as JSON and summarised in the monthly GPU price report.
How this price is calculated
All prices on this page are normalized to a per-GPU hourly rate using each offer's authoritative GPU count, so that raw price is divided by the number of GPUs it actually covers; some offers report price as already per-GPU, so those are used as-listed. An offer with a missing, zero, or invalid GPU count is excluded entirely rather than published at a guessed rate.
No offers were excluded from this snapshot for a missing or invalid price. No offers were dropped as price outliers in this snapshot.
Only the cheapest qualifying offer per provider is shown in the table above. This page regenerates at most once per hour.
AMD MI300X vs H100: how do they compare?
- VRAM: AMD MI300X 192GB HBM3 vs H100 80GB HBM3
- Memory bandwidth: 5.3 TB/s vs 3.35 TB/s
- Dense FP8 tensor throughput: 2,614.9 TFLOPS vs 1,979 TFLOPS
On Aquanode right now, H100 starts at $2.19/GPU/hr against AMD MI300X's $2.63/GPU/hr, about 17% less.
AMD MI300X vs B200: how do they compare?
- VRAM: AMD MI300X 192GB HBM3 vs B200 180GB HBM3e
- Memory bandwidth: 5.3 TB/s vs 8 TB/s
- Dense FP8 tensor throughput: 2,614.9 TFLOPS vs 4,500 TFLOPS
On Aquanode right now, AMD MI300X starts at $2.63/GPU/hr against B200's $8.79/GPU/hr, about 70% less.
Compare the AMD MI300X with other GPUs
Compare AMD MI300X with
- AMD MI300X vs B200
- AMD MI300X vs H200
- AMD MI300X vs H100
- A100 vs AMD MI300X
- AMD MI300X vs DGX A100
- AMD MI300X vs V100
- AMD MI300X vs L40S
- AMD MI300X vs L40
- A40 vs AMD MI300X
- AMD MI300X vs RTX A6000
- AMD MI300X vs RTX A5000
- AMD MI300X vs RTX A4000
- AMD MI300X vs L4
- AMD MI300X vs T4
- AMD MI300X vs RTX 6000 Ada
- AMD MI300X vs RTX PRO 6000
- AMD MI300X vs RTX PRO 6000 WS
- AMD MI300X vs RTX PRO 6000 SE
- AMD MI300X vs RTX PRO 5000
- AMD MI300X vs RTX 5090
- AMD MI300X vs RTX 5080
- AMD MI300X vs RTX 5070 Ti
- AMD MI300X vs RTX 5070
- AMD MI300X vs RTX 5060 Ti
- AMD MI300X vs RTX 4090
- AMD MI300X vs RTX 4080 Super
- AMD MI300X vs RTX 4080
- AMD MI300X vs RTX 4070 Ti
- AMD MI300X vs RTX 4070 Super
- AMD MI300X vs RTX 4070
- AMD MI300X vs RTX 4060 Ti
- AMD MI300X vs RTX 3090
- AMD MI300X vs RTX 3080
- AMD MI300X vs RTX 3070
- AMD MI300X vs RTX 3060
- AMD MI300X vs B300
- AMD MI300X vs AMD MI355X
- AMD MI300X vs AMD MI325X
- AMD MI300X vs H100 NVL
- AMD MI300X vs RTX PRO 4500
- AMD MI300X vs RTX PRO 4500 SE
- AMD MI300X vs RTX PRO 4000
- AMD MI300X vs RTX 5880 Ada
- AMD MI300X vs RTX 5000 Ada
- AMD MI300X vs RTX 4000 Ada
- AMD MI300X vs RTX 4000 SFF Ada
- AMD MI300X vs RTX 2000 Ada
- AMD MI300X vs RTX 4070 Ti Super
- AMD MI300X vs RTX A4500
- AMD MI300X vs RTX 2080 Ti
- AMD MI300X vs RTX 3080 Ti
- AMD MI300X vs RTX 3070 Ti
- AMD MI300X vs RTX 3060 Ti
- AMD MI300X vs GTX 1080 Ti
- AMD MI300X vs RTX 4060
- AMD MI300X vs RTX 5060
- AMD MI300X vs GTX 1660 Super
- AMD MI300X vs RTX 2060
- AMD MI300X vs Radeon RX 9070 XT
- AMD MI300X vs Radeon RX 6700 XT
- AMD MI300X vs RTX 6000
- A16 vs AMD MI300X
- AMD MI300X vs P40
- AMD MI300X vs P4
- AMD MI300X vs Quadro P2000
- AMD MI300X vs Quadro M4000
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.
Rent a AMD MI300X on Aquanode
- On-demand instances from $2.63/GPU/hr, billed by the provider's own terms, with no hardware procurement or long-term commitment.
- 10 live offers across 2 regions today.
- Set a price/availability alert above to hear the moment a cheaper or newly-available AMD MI300X offer appears.
- Compare every AMD MI300X offer side by side, or browse the full multi-provider GPU marketplace.
Good for
192GB on a single card (more than double an H100's 80GB), so it fits a 70B model in FP16 with room to spare and holds larger models without splitting across GPUs, at a memory bandwidth (5.3 TB/s) ahead of H100's 3.35 TB/s.
Not good for
Runs on AMD's ROCm software stack rather than CUDA. Most inference/training frameworks support it, but some CUDA-only tooling and custom kernels need a ROCm port or don't run at all. Worth checking your stack's ROCm support before committing to it.
AMD MI300X FAQs
How much VRAM does the AMD MI300X have?
The AMD MI300X has 192GB HBM3, with 5.3 TB/s of peak memory bandwidth.
What is the AMD AMD MI300X?
The AMD AMD MI300X is a GPU released in 2023, with 192GB HBM3 of memory and a 750W power envelope. See the full spec table above for interconnect, form factor and tensor-throughput details.
How much does it cost to rent a AMD MI300X?
Live AMD MI300X rental prices currently range from $2.63 to $3.29 per GPU per hour, with a median of $2.63 per GPU per hour. Out of capacity right now.
What's the cheapest AMD MI300X rate?
The lowest current AMD MI300X rate on Aquanode is $2.63 per GPU per hour in . Out of capacity right now.
How does the AMD MI300X compare to the H100?
VRAM: AMD MI300X 192GB HBM3 vs H100 80GB HBM3 Memory bandwidth: 5.3 TB/s vs 3.35 TB/s Dense FP8 tensor throughput: 2,614.9 TFLOPS vs 1,979 TFLOPS On Aquanode right now, H100 starts at $2.19/GPU/hr against AMD MI300X's $2.63/GPU/hr, about 17% less. See the full AMD MI300X vs H100 comparison for a shared-provider price breakdown.
How does the AMD MI300X compare to the B200?
VRAM: AMD MI300X 192GB HBM3 vs B200 180GB HBM3e Memory bandwidth: 5.3 TB/s vs 8 TB/s Dense FP8 tensor throughput: 2,614.9 TFLOPS vs 4,500 TFLOPS On Aquanode right now, AMD MI300X starts at $2.63/GPU/hr against B200's $8.79/GPU/hr, about 70% less. See the full AMD MI300X vs B200 comparison for a shared-provider price breakdown.
What is the AMD MI300X good for?
192GB on a single card (more than double an H100's 80GB), so it fits a 70B model in FP16 with room to spare and holds larger models without splitting across GPUs, at a memory bandwidth (5.3 TB/s) ahead of H100's 3.35 TB/s.
What are the AMD MI300X's limitations?
Runs on AMD's ROCm software stack rather than CUDA. Most inference/training frameworks support it, but some CUDA-only tooling and custom kernels need a ROCm port or don't run at all. Worth checking your stack's ROCm support before committing to it.
Is renting cheaper than buying?
Renting avoids the upfront hardware cost and lets you match spend to actual usage. A rented AMD MI300X at $2.63/hr only costs money while it's running, whereas buying ties up capital in hardware that keeps depreciating whether it's in use or not. Which is cheaper depends on how continuously you'd run it; short or bursty workloads usually favor renting. Out of capacity right now.
How is the AMD MI300X price calculated?
All prices are normalized to a per-GPU hourly rate using each offer's authoritative GPU count, which the raw price is divided by; some offers report price as already per-GPU. Offers whose price can't be safely normalized, or whose rate is an extreme outlier against the rest of the market, are excluded.
AMD MI300X price by region
Related guides
Other models in the same generation, then the rest of the GPU index.
