NVIDIA · Blackwell · 2025
NVIDIA RTX 5090. Rental by the hour, on demand.
From $0.72/hr per GPU on an 8-GPU machine, pay as you go with no commitment. A Blackwell desktop card for AI prototyping, with 32 GB of GDDR7.
RTX 5090 on demand
on an 8-GPU machine
- Availability
- Available now · 5 regions
- By region and size
- $0.72 to $1.84/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
32 GB of GDDR7. At 1.8 TB/s.
- Memory
- 32 GB
- GDDR7
- Bandwidth
- 1.8 TB/s
- Memory to compute
- FP16
- 209.5
- non-tensor TFLOPS
- Per node
- 2 GPUs
- PCIe, 575 W each
Price across clouds
NVIDIA RTX 5090, 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 5090. Each one opens the launch form, sized.
Launch it
Ask your agent for an RTX 5090. 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 5090 on Runcrate called my-gpu, copy train.py to it and run it.
create_instance { name: "my-gpu", gpu: "RTX5090" }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 5090 compares. Against its closest alternatives.
Benchmarks
The RTX 5090, measured. Real runs, dated, with what they cost.
- Speech to text
Whisper large-v3 audio hours per dollar (run 2026-10-28)
- Text embeddings
BGE embedding throughput per dollar (run 2026-10-29)
- Training and fine-tuning
torch.compile vs eager for training (run 2026-10-29)
- Image generation
SDXL images per dollar on four GPUs (run 2026-10-27)
Alternatives
Alternatives to the RTX 5090. Same account, different price and performance.
| GPU | Memory · FP16 non-tensor | Stands out | Price gap | From |
|---|---|---|---|---|
| NVIDIA RTX 5090This GPU | 32 GB GDDR7 · 209.5 TFLOPS | $0.72/hr | ||
| AMD MI210 | 64 GB HBM2e | +32 GB memory | Not in stock right now | |
| NVIDIA RTX 4090 | 24 GB GDDR6X · 165.2 TFLOPS | -8 GB memory | -19% | $0.58/hr |
| NVIDIA RTX A6000 | 48 GB GDDR6 | +16 GB memory | -21% | $0.57/hr |
| NVIDIA A40 | 48 GB GDDR6 | +16 GB memory | +269% | $2.66/hr |
Full specs
NVIDIA RTX 5090 specifications.
Memory
- Memory
- 32 GB
- Type
- GDDR7
- Bandwidth
- 1.8 TB/s
Compute
- FP32
- 104.8 TFLOPS
- FP16 non-tensor
- 209.5 TFLOPS
- INT8
- 838 TOPS
- CUDA cores
- 21,760
Power and form
- TDP
- 575 W
- Form factor
- Consumer
- Architecture
- Blackwell
- Released
- 2025
Node
- GPUs per node
- 2
- Interconnect
- PCIe
- Maker
- NVIDIA
Questions
Renting the RTX 5090. What buyers ask first.
Other GPUs
Browse every GPU by architecture.
NVIDIA Blackwell
NVIDIA Hopper
NVIDIA Ampere
NVIDIA Ada Lovelace
NVIDIA Volta