Rent an RTX 4090 by the hour
On PowerPod an RTX 4090 costs from USD 0.54 an hour today, billed by the second, on 38 verified hosts. The price you see when you rent is locked for that instance.
What an RTX 4090 is good for
GeForce RTX 40 series: excellent value for fine-tuning, image generation and inference. With 24 GB of memory it is a good fit for:
- QLoRA fine-tuning of 7B–14B models such as Llama, Qwen and Gemma
- Serving 7B–14B models at full speed, or 30B-class models with 4-bit weights
- SDXL and FLUX image generation, and training image LoRAs
- Whisper large-v3 transcription in large batches
What common jobs cost on an RTX 4090
Estimated from typical run times and today’s lowest price. Your own job may run faster or slower.
| Job | Typical time | Cost today |
|---|---|---|
| Fine-tune a 7B model | 3 h | USD 1.62 |
| Generate 1,000 images | 0.7 h | USD 0.38 |
| Transcribe 100 hours of audio | 3 h | USD 1.62 |
| Serve a chat model for an evening | 4 h | USD 2.16 |
| A working day in JupyterLab | 8 h | USD 4.32 |
Price over time
Where the hosts are
RTX 4090 hosts are listed today in 15 countries: Canada, Denmark, Spain, United Kingdom, Hong Kong SAR China, Hungary, Ireland, Japan, South Korea, Macao SAR China, Norway, Romania, Sweden, Taiwan, United States. 1 of them is in a datacenter. Each offer shows its own measured reliability before you rent.
Typical machine
Median across the hosts listed today; each offer shows its exact machine before you rent.
| Memory bandwidth | 879 GB/s |
| Compute | 81.4 TFLOPS |
| Highest CUDA version | 13.2 |
| CPU cores for you | 24 |
| System memory for you | 16 GB |
| Download speed | 926 Mbps |
Questions
How much does it cost to rent an RTX 4090?
On PowerPod an RTX 4090 costs from USD 0.54 an hour today, billed by the second, on 38 verified hosts. The price you see when you rent is locked for that instance.
How much memory does the RTX 4090 have?
24 GB of GPU memory (some hosts offer 48 GB).
What can I run on an RTX 4090?
QLoRA fine-tuning of 7B–14B models such as Llama, Qwen and Gemma. Serving 7B–14B models at full speed, or 30B-class models with 4-bit weights. SDXL and FLUX image generation, and training image LoRAs. Whisper large-v3 transcription in large batches.
Can I pay in rupees?
Yes. Accounts in Nepal are billed in Nepali rupees and accounts in India in Indian rupees, paid with local methods; everyone else pays in US dollars. Prices can be shown in about 160 currencies.