AMD Instinct MI300X cloud GPU rental
Rent an AMD Instinct MI300X cloud GPU from $2.58 per hour with 192 GB of HBM3 at 5.3 TB/s of memory bandwidth. FastInfra bills pods per minute with no commitment, so a one-hour experiment costs $2.58 and you stop paying the moment you delete the pod. Run 1 to 8 GPUs per pod, connect over SSH or JupyterLab, and see the exact price before you deploy.
Prices synced November 3, 2026 · billed per minute
MI300X specs and pricing
The AMD Instinct MI300X is a datacenter CDNA 3 GPU with 192 GB of HBM3 at 5.3 TB/s of memory bandwidth. On FastInfra it is a good fit for frontier-scale training, 200B+ inference, long-context serving. As a rule of thumb, one MI300X holds the weights of a model of about 76B parameters at 16-bit precision, or about 300B with 4-bit quantization, after leaving room for the KV cache.
The MI300X runs on the Secure tier in vetted datacenters at $2.58 per GPU-hour.
MI300X price per hour by GPU count
Live FastInfra prices per pod-hour. Pods bill per minute, so you pay only for the minutes a pod exists.
| Pod size | Total VRAM | Secure / hr | Month (730 h, lowest) |
|---|---|---|---|
| 1× MI300X | 192 GB | $2.58 | $1,885 |
| 2× MI300X | 384 GB | $5.16 | $3,770 |
| 4× MI300X | 768 GB | $10.33 | $7,539 |
| 8× MI300X | 1536 GB | $20.66 | $15,079 |
What an MI300X costs over time
One GPU at the lowest on-demand rate ($2.58/hr).
| Duration | On-demand |
|---|---|
| 1 hour | $2.58 |
| 8-hour working day | $20.66 |
| 24 hours | $61.97 |
| 1 week (168 hours) | $434 |
| 1 month (730 hours) | $1,885 |
Which LLMs fit on the MI300X?
GPUs needed to hold the model weights with about 20% of VRAM left for the KV cache. Rule of thumb only: long contexts and big batches need more.
| Model size | Examples | FP16 / BF16 | 8-bit | 4-bit |
|---|---|---|---|---|
| 7–8B | Llama 3.1 8B, Qwen3 8B, Mistral 7B | 1 GPU | 1 GPU | 1 GPU |
| 12–14B | Gemma 3 12B, Qwen3 14B, Phi-4 | 1 GPU | 1 GPU | 1 GPU |
| 27–32B | Gemma 3 27B, Qwen3 32B | 1 GPU | 1 GPU | 1 GPU |
| 70B | Llama 3.3 70B | 1 GPU | 1 GPU | 1 GPU |
| 120B | GPT-OSS 120B | 2 GPUs | 1 GPU | 1 GPU |
| 235B | Qwen3 235B-A22B | 4 GPUs | 2 GPUs | 1 GPU |
| 405B | Llama 3.1 405B | 8 GPUs | 4 GPUs | 2 GPUs |
| 671B | DeepSeek V3 / R1 | Needs more GPUs | 8 GPUs | 4 GPUs |
Best for
Frontier-scale training, 200B+ inference, long-context serving.
About the CDNA 3 architecture
CDNA 3 is AMD's datacenter architecture. The MI300X pairs 192 GB of HBM3 with 5.3 TB/s of bandwidth, so a single GPU can hold models that need two NVIDIA 80 GB cards. PyTorch, vLLM and SGLang support it through ROCm; CUDA-only code and custom kernels need porting with HIP.
Only need inference?
If you want to call an open model such as Llama, Qwen, DeepSeek or GPT-OSS rather than manage a server, the pay-per-token inference API is often cheaper than keeping a GPU busy around the clock. It is OpenAI-compatible and uses the same account and wallet as your pods. Browse every model with live prices.
MI300X vs similar GPUs
Live lowest on-demand price per GPU-hour, and price per GB of VRAM.
| GPU | VRAM | Architecture | From / hr | Per GB-hr |
|---|---|---|---|---|
| MI300X (this page) | 192 GB | CDNA 3 | $2.58 | $0.013 |
| NVIDIA B300 | 288 GB | Blackwell | $7.50 | $0.026 |
| NVIDIA B200 | 180 GB | Blackwell | $6.46 | $0.036 |
| NVIDIA H200 NVL | 143 GB | Hopper | $4.09 | $0.029 |
| NVIDIA H200 SXM | 141 GB | Hopper | $3.88 | $0.028 |
| NVIDIA RTX PRO 6000 Max-Q | 96 GB | Blackwell | $1.77 | $0.018 |
| NVIDIA RTX PRO 6000 | 96 GB | Blackwell | $1.83 | $0.019 |
MI300X rental — frequently asked questions
How much does it cost to rent an AMD Instinct MI300X per hour?
$2.58 per GPU-hour on demand. Billing is per minute while the pod exists, so 24/7 use for a 730-hour month costs about $1,885 per GPU at the lowest on-demand rate. Prices sync from live capacity and are shown before you deploy.
How much VRAM does the AMD Instinct MI300X have?
192 GB of HBM3 at 5.3 TB/s of memory bandwidth. The chip is rated at up to 750 W. A pod with 8 GPUs gives 1,536 GB of combined VRAM for tensor- or pipeline-parallel workloads.
What size LLM can I run on an AMD Instinct MI300X?
Roughly 76B parameters at FP16 or BF16, 150B at 8-bit and 300B at 4-bit on a single GPU, reserving about 20% of VRAM for the KV cache and activations. Mixture-of-experts models need memory for all experts, not only the active ones. Long contexts and large batches need more headroom.
Is the AMD Instinct MI300X billed per hour or per minute?
Per minute. The hourly rate is divided by 60 and charged for every minute the pod exists. Attached volumes bill per GB-month and continue while a pod is stopped; delete the pod to stop all charges. There is no minimum term or reservation fee.
How do I rent an AMD Instinct MI300X on FastInfra?
Create an account, add prepaid credit, open the GPU configurator and choose the MI300X, the tier, the GPU count, a container template and storage. The configurator shows the hourly and monthly price, then deploys a pod you reach over SSH or JupyterLab within minutes.
Can I rent multiple MI300X GPUs in one pod?
Yes, up to 8 per pod. The price scales linearly: 8 GPUs cost $20.66 per hour on Secure.
Deploy an MI300X in minutes
Price the exact pod, storage and template in the configurator, add credit, and connect over SSH or JupyterLab. Billed per minute, no commitment.