NVIDIA · Ada Lovelace · 2022
NVIDIA RTX 4090. Rental by the hour, on demand.
From $0.58/hr per GPU on an 8-GPU machine, on demand, or prepay an 8-GPU machine at $0.56/GPU-hr for 1 month, $0.54/GPU-hr for 3 months.
RTX 4090 on demand
on an 8-GPU machine
- Availability
- Available now · 3 regions
- By region and size
- $0.58 to $0.79/hr
- Prepaid
- 1 mo $0.56/hr3 mo $0.54/hr6 mo $0.52/hr12 mo $0.50/hr
- Billing
- By the second on most GPUs, a few by the minute
- 8+ GPUs or a month
- Launch any size yourself, or prepay a machine for 1 to 12 months at a discount. Custom quotes on request.
The card
24 GB of GDDR6X. At 1.0 TB/s.
- Memory
- 24 GB
- GDDR6X
- Bandwidth
- 1.0 TB/s
- Memory to compute
- FP16
- 165.2
- non-tensor TFLOPS
- Per node
- 2 GPUs
- PCIe, 450 W each
Price across clouds
NVIDIA RTX 4090, cloud by cloud. Sorted by price.
On-demand rates per GPU-hour from each cloud's public price page on 8 Nov 2026: RunPod's Secure Cloud, and Lambda's lowest per-GPU rate across its cluster sizes. Runcrate's row is today's live starting rate.
Workloads
What teams run on the RTX 4090. Each one opens the launch form, sized.
Launch it
Ask your agent for an RTX 4090. Or read the call it makes.
Claude Code or Codex runs it for you: it starts the instance, copies your files over and runs the job. The SDKs make the same request.
Connect your agent first
Add Runcrate's MCP server to yourself: run curl -fsSL "https://runcrate.ai/mcp.md" and follow it.
Start an RTX 4090 on Runcrate called my-gpu, copy train.py to it and run it.
create_instance { name: "my-gpu", gpu: "RTX4090" }instance_status { id: "my-gpu" }file_upload { id: "my-gpu", content: "<train.py>", path: "/root/workspace/train.py" }ssh_execute { id: "my-gpu", command: "nohup python /root/workspace/train.py > train.log 2>&1 &" }Spec sheet
How the RTX 4090 compares. Against its closest alternatives.
Alternatives
Alternatives to the RTX 4090. Same account, different price and performance.
| GPU | Memory · FP16 non-tensor | Stands out | Price gap | From |
|---|---|---|---|---|
| NVIDIA RTX 4090This GPU | 24 GB GDDR6X · 165.2 TFLOPS | $0.58/hr | ||
| NVIDIA RTX A6000 | 48 GB GDDR6 | +24 GB memory | -2% | $0.57/hr |
| NVIDIA A40 | 48 GB GDDR6 | +24 GB memory | +359% | $2.66/hr |
| AMD MI210 | 64 GB HBM2e | +40 GB memory | Not in stock right now | |
| NVIDIA V100 | 32 GB HBM2 | +8 GB memory | +478% | $3.35/hr |
Full specs
NVIDIA RTX 4090 specifications.
Memory
- Memory
- 24 GB
- Type
- GDDR6X
- Bandwidth
- 1.0 TB/s
Compute
- FP32
- 82.6 TFLOPS
- FP16 non-tensor
- 165.2 TFLOPS
- INT8
- 661 TOPS
- CUDA cores
- 16,384
Power and form
- TDP
- 450 W
- Form factor
- Consumer
- Architecture
- Ada Lovelace
- Released
- 2022
Node
- GPUs per node
- 2
- Interconnect
- PCIe
- Maker
- NVIDIA
Questions
Renting the RTX 4090. What buyers ask first.
Other GPUs
Browse every GPU by architecture.
NVIDIA Blackwell
NVIDIA Hopper
NVIDIA Ampere
NVIDIA Ada Lovelace
NVIDIA Volta