New B200 spot capacity is live in US East from $1.69 per GPU-hour. See availability
NVIDIA · Ampere · Consumer GPU

RTX 3090 Ti from $0.06 per GPU-hour

RTX 3090 Ti is a consumer board rented by the hour with 24 GB of GDDR6X, built on the Ampere architecture in 2022. It holds models up to about 9B at FP16, and 30B-class models quantised to 4 bits. Memory bandwidth is 1,008 GB/s, which is what sets token throughput for inference.

Specifications from the vendor datasheet · prices updated recently

Price per GPU-hour 24 available
SpotReclaimed with 2 min notice · 10–15% over 30 days $0.06$0.0010 / min
On-demandRuns until you stop it $0.09−33% on spot
ReservedFixed rate, 1 to 12 months $0.06monthly term

Billed per started minute. 96 GB of GPU memory on a 4× node. Persistent storage $0.08 per GB-month. No egress fees.

GPU memory
24GBGDDR6X
Bandwidth
1,008GB/sSets token throughput on inference
Node sizes
1× · 2× · 4×Same price per GPU at every size
Reclaim rate
10–15%Spot instances reclaimed over 30 days
Node sizes

What you get with each RTX 3090 Ti node.

All prices
Node GPU memory vCPU RAM Local NVMe Spot On-demand Action
1× RTX 3090 Ti 24 GB 8 48 GB 500 GB $0.06/h $0.09/h Configure
2× RTX 3090 Ti 48 GB 16 96 GB 1 TB $0.12/h $0.18/h Configure
4× RTX 3090 Ti 96 GB 32 192 GB 2 TB $0.24/h $0.36/h Configure

GPU-to-GPU link: PCIe Gen4 x16. Local NVMe is scratch space wiped when the instance ends; keep anything you need on a persistent disk.

Availability

24 RTX 3090 Ti GPUs free right now.

Straight from the capacity pool the console books against. Prices are identical in every region — pick the one closest to your data.

US WestOregon, US 24
Tier III100 Gbps per nodeus-west

A region with zero free GPUs still accepts reserved capacity requests — we hold hardware for a term rather than sell what is already taken.

Memory planning

What fits in 24 GB — and what it costs per hour.

Weights need about 2 GB per billion parameters at FP16 / BF16, 0.5 GB per billion parameters at INT4. We keep 20% of the memory free for activations, the KV cache and the CUDA context, then take the smallest node that still fits.

Open model Parameters FP16 / BF16node · spot price INT4node · spot price
Mistral 7BAssistants, RAG, classification 7.2B $0.06/h $0.06/h
Llama 3.1 8BAssistants, agents, fine-tuning 8B $0.06/h $0.06/h
Gemma 2 27BHigher-quality assistants 27B $0.24/h $0.06/h
Qwen2.5 32BCode and reasoning 32B $0.24/h $0.06/h
Mixtral 8x7BMixture of experts, high throughput 46.7B over 4× $0.12/h
Llama 3.3 70BThe common production baseline 70B over 4× $0.12/h
Qwen2.5 72BMultilingual, long context 72B over 4× $0.12/h
Mixtral 8x22BMixture of experts, large capacity 141B over 4× $0.24/h
Llama 3.1 405BLargest widely used open model 405B over 4× over 4×
This is an estimate for weights, not a benchmark.

Real usage depends on context length, batch size and the serving engine: a long context can add tens of gigabytes of KV cache. Treat the table as the floor, not the ceiling — and if a job is close to the limit, take the next node size or quantise one step further.

Value

What a dollar buys on an RTX 3090 Ti.

Ranked 3 of 14 consumer models in this catalogue on cost per gigabyte of GPU memory.

VRAM per dollar
400 GB

GPU memory you get for $1.00 of spot time, per hour.

Memory bandwidth per dollar
16,800 GB/s

Bandwidth decides token throughput far more often than raw FLOPS.

Cost per FP32 TFLOP-hour
$0.00150

Spot price divided by published FP32 throughput.

Cost of a 24-hour run
$1.44

One GPU on spot, billed per minute, storage excluded.

Against the rest of the market

5 providers · September 2026
Median on-demand elsewhere
$0.19Our on-demand is 53% below it
Cheapest on-demand seen
$0.16Our on-demand $0.09 is 44% below it
Cheapest spot seen
$0.10Our spot $0.06 is 40% below it
24 hours on one GPU
$1.44 on spot$2.16 on-demand · $3.84 at the cheapest rate found elsewhere

Index built from published prices for the same GPU across 137 providers (public price index (getdeploying.com)), September 2026. How the guarantee is enforced.

Good for

What people run on an RTX 3090 Ti.

Fine-tuning small models

LoRA on 7B–13B models, and full fine-tuning below 3B.

Image and video generation

Diffusion pipelines run entirely in memory at this capacity.

Multi-GPU jobs over NVLink

GPU-to-GPU traffic stays off the PCIe bus, which is what makes tensor and pipeline parallelism worth it.

Anything that can checkpoint

On the spot tier this model is 33% below its own on-demand price, with a 2-minute notice before a reclaim.

Specifications

NVIDIA RTX 3090 Ti, as published by NVIDIA.

Taken from the vendor datasheet. Where a figure is not published, the row is absent rather than estimated.

Vendor
NVIDIA
Architecture
Ampere (2022)
Segment
Consumer
Memory
24 GB GDDR6X
Memory bandwidth
1,008 GB/s
FP32
40 TFLOPS
CUDA cores
10,752
Board power
450 W
Form factor
PCIe
Interconnect
PCIe Gen4 x16
Multi-instance (MIG)
Not available
8-bit float (FP8)
Not supported — use FP16 or integer quantisation

Source: images.nvidia.com · www.nvidia.com · verified against the vendor document

Software that runs on it

8 templates

Pre-built environments, pulled on the node before you land on it. Or bring any OCI image from a public or private registry.

PyTorch 2.5 · CUDA 12.4CUDA 12.4 basevLLM inferenceJupyterLabComfyUIOllamaTensorFlow 2.17Ubuntu 22.04 (driver only)

All templates

On the spot tier

10–15% reclaimed · 30 d
  • 2 minutes of notice on the metadata endpoint, a webhook and the console
  • The instance is stopped, not deleted — disk and IP stay attached
  • Auto-relaunch on the next free RTX 3090 Ti, or switch the same disk to on-demand
  • You save $0.72 a day per GPU against on-demand

How interruptions work

FAQ

RTX 3090 Ti — the questions that come up

How much does an RTX 3090 Ti cost per hour?

Spot is $0.06 per GPU-hour and on-demand is $0.09, both billed per minute — every started minute costs the hourly price divided by 60, so $0.0010 on spot. Reserved capacity is $0.06 per GPU-hour on a monthly commitment. Storage and egress are not included in that rate: persistent disks are $0.08 per GB-month and there are no egress fees.

How many GPUs can I put in one instance?

Node sizes are 1×, 2×, 4× — up to 4 RTX 3090 Ti GPUs in a single virtual machine, with CPU, RAM and local NVMe scaled with the GPU count. The price per GPU is identical at every node size.

Can it run a 70B model?

Not on a single node of this model: at FP16 a 70B model needs about 140 GB of memory for the weights alone, more than 4× 24 GB leaves free. Quantised to 4 bits it needs about 35 GB — check the memory table above — or pick a model with more memory per GPU.

Which regions have it?

Available now in US West (Oregon). Prices are identical in every region; pick the one closest to your data.

How often is a spot RTX 3090 Ti reclaimed?

Over the trailing 30 days, 10–15% of spot instances on this model were reclaimed. You get a 2-minute notice on the metadata endpoint, through a webhook and in the console; the instance is then stopped, never deleted, and the persistent disk stays attached.

Anything else about this model? Ask an engineer — the same people run the nodes.

Get started

An RTX 3090 Ti for $0.06 an hour, running in under 60 seconds.

Pay as you go — no contracts, no minimum commitment. Add credit, launch, stop whenever you want.

Billed per minute from the moment the instance is reachable. Minimum credit $40, no subscription. 24 RTX 3090 Ti GPUs available right now.