RTX 4090 cloud rental in 2026: price per hour, what fits in 24 GB, and when to pick a 5090 or a 48 GB card
Rent one RTX 4090 on Lium from $0.34 per hour on Spot or $0.35 on Secure, or a whole 8-GPU node from $0.55 per GPU-hour. 24 GB holds an 8B model in 16-bit or a 32B model in 4-bit. Prices checked 18 November 2026.
Lium team3 min read
rtx-4090
inference
pricing
buyer-guide

Short answer: one RTX 4090 on Lium costs from $0.34 per hour on the Spot tier and from $0.35 on the Secure tier. A whole 8-GPU node costs from $0.55 per GPU-hour, or $4.40 per hour for all eight. Rent a 4090 when your model fits in 24 GB. Rent an RTX 5090 when it needs up to 32 GB, and a 48 GB card when it needs more.
(Lium prices from the public nodes feed, generated 18 November 2026 at 09:39 UTC. Lium is a marketplace where providers set the price, so check Pricing for today's numbers and Stats for how many GPUs are listed and rented.)
RTX 4090 price per hour on Lium
These are the cheapest asks for nodes you could rent at that moment. Some providers only rent a multi-GPU node as a whole, and at this read those whole nodes asked more per GPU than the cheapest single GPUs. If you need one or two GPUs, rent them singly.
Spot can be reclaimed, so save checkpoints and results off the pod. Lium bills per second, so a 10-minute test costs 10 minutes.
RTX 4090 specs
From NVIDIA's RTX 4090 page (read 18 November 2026):
- Memory: 24 GB GDDR6X on a 384-bit interface.
- CUDA cores: 16,384.
- Architecture: Ada Lovelace, with 4th-generation Tensor Cores.
- NVLink: no. Multi-GPU 4090 nodes talk over PCIe.
What fits in 24 GB
The weights take most of the memory. Each parameter is 2 bytes in 16-bit, 1 byte in 8-bit and about half a byte in 4-bit (see Hugging Face's quantization overview, read 18 November 2026). The KV cache for your context and batch comes on top.
An 8B model in 16-bit leaves about 8 GB for the KV cache, which is enough for moderate contexts and small batches. A 32B model in 4-bit fits the same way.
RTX 4090 vs RTX 5090 vs 48 GB cards on Lium
Same nodes feed, same read time. L40S (48 GB) was listed from $0.38 per GPU-hour, but only in nodes of 2 or more GPUs, so the smallest L40S rental was $0.76 per hour.
- Pick a 4090 when your model fits in 24 GB with room for its context, for example an 8B model in 16-bit or a 32B model in 4-bit. It is the cheapest per hour.
- Pick a 5090 when the weights need more than 24 GB but fit in 32 GB, such as a 14B model in 16-bit. See RTX 5090 cloud rental in 2026.
- Pick a 48 GB card such as the RTX A6000 when you need more memory on one GPU, for example a 32B model in 8-bit. It is an older generation (Ampere) than the 4090, so check that memory is your limit before you switch.
How to rent an RTX 4090
Install the CLI and sign in once:
bash
curl -fsSL https://lium.io/install.sh | bash
lium initRent one RTX 4090 for up to 2 hours:
bash
lium up --gpu RTX4090 --name my4090 --ttl 2h -yStop it when you are done:
bash
lium rm my4090 -yTo serve a model on it, follow Serve an open LLM with vLLM on a single Lium GPU. For a model-size guide across all GPUs, see Best GPU cloud for open-source LLM inference.
Prices change hourly on a marketplace. Check Pricing before you rent.
Run it on Lium
Get new posts by e-mail: Subscribe to the Lium newsletter


