AMD MI300X

What size AI models can the MI300X run?

Model	Parameters	4-bitINT4 / FP4	8-bitFP8 / INT8	16-bitFP16 / BF16
Qwen3.8-27B	27B	1 GPU 18 GB VRAM · $2,102/mo	1 GPU 30 GB VRAM · $2,102/mo	1 GPU 58 GB VRAM · $2,102/mo
Gemma 4 31B	30.7B	1 GPU 22 GB VRAM · $2,102/mo	1 GPU 35 GB VRAM · $2,102/mo	1 GPU 69 GB VRAM · $2,102/mo
GPT-OSS-120B	117B 5.1B active	1 GPU 67 GB VRAM · $2,102/mo	1 GPU 120 GB VRAM · $2,102/mo	2 GPUs 237 GB VRAM · $4,205/mo
DeepSeek V4 Flash	284B 13B active	1 GPU 160 GB VRAM · $2,102/mo	2 GPUs 287 GB VRAM · $4,205/mo	4 GPUs 573 GB VRAM · $8,410/mo
GLM-5.3-Flash	320B 18B active	2 GPUs 178 GB VRAM · $4,205/mo	2 GPUs 322 GB VRAM · $4,205/mo	4 GPUs 642 GB VRAM · $8,410/mo
MiniMax-M3	428B 23B active	2 GPUs 239 GB VRAM · $4,205/mo	4 GPUs 432 GB VRAM · $8,410/mo	8 GPUs 862 GB VRAM · $16,819/mo
Kimi K3	2.8T 104B active	9 GPUs 1,544 GB VRAM	17 GPUs 2,804 GB VRAM	33 GPUs 5,605 GB VRAM

Estimates based on the median on-demand rate. Memory is weights plus FP8 KV cache at 32K context per request (FP16/BF16 in the 16-bit column). GPU counts assume 90% of advertised VRAM is usable. No guarantee of runtime support, usable performance, or that a matching quantized build exists. How we estimate costs.

Frequently Asked Questions

How much does the MI300X cost per hour?

As of September 7, 2026, the median on-demand price is $2.92 per GPU per hour across 6 providers with a priced on-demand config.

Billing type	Configs	Median / GPU / hr	Cheapest
On-demand	14	$2.92	$2.39 (in stock, Runpod)
Reserved	23	$3.39	$1.63 (12 mo)
Spot	3		$1.60
Custom contract	1		$1.71

No median for spot and custom-contract: we only show this when at least 3 providers list the GPU on that billing type. Custom-contract pricing is negotiated per deal and usually not published.

Who has the cheapest MI300X?

As of September 7, 2026, the cheapest verified in-stock on-demand MI300X tracked by GetDeploying is $2.39 per GPU per hour from Runpod.

How much does the MI300X cost per month?

As of September 7, 2026, at 720 hours per month, one MI300X costs an estimated $2,102 at the median on-demand price. Cheapest verified in stock: $1,721 per month on-demand.

Where can I rent an MI300X?

GetDeploying currently tracks MI300X configs from 10 providers. Runpod has verified in-stock on-demand configs. See the full price comparison above for every provider and config.

Are MI300X prices going up or down?

As of September 7, 2026, the median on-demand price has fallen about 7% over the past 90 days to $2.92 per GPU per hour, though it is about the same as a year ago.

How many MI300X GPUs do I need?

One MI300X runs models up to roughly 303B parameters at 4-bit quantization or 81B at 16-bit, assuming a 32K context. Larger models run across multiple GPUs: the model table above shows the estimated memory and GPU count for popular open-weight LLMs.

Why choose the MI300X?

Comes with 192GB HBM3 - the largest memory of any single GPU. 5.3 TB/s memory bandwidth. Strong ROCm ecosystem for PyTorch workloads. Competitive pricing against H100 for inference.

When is the MI300X not a good fit?

ROCm software ecosystem, while rapidly improving, is narrower than CUDA. Some frameworks and libraries may have limited AMD support. Multi-node training tooling is less mature than Nvidia's.

Alternatives to AMD MI300X

• 

  Nvidia H100

  80 GB HBM3 · 37% less bandwidth than the MI300X

  From $1.73 / GPU / hr Compare
• 

  Nvidia A100

  80 GB HBM2e · 63% less bandwidth than the MI300X

  From $0.47 / GPU / hr Compare
• 

  AMD MI325X

  256 GB HBM3e · released Q4 2024

  From $2.25 / GPU / hr Compare