NVIDIA RTX PRO 4000 Blackwell SFF 24GB vs NVIDIA GeForce RTX 5090 32GB

The GeForce RTX 5090 delivers 4.4x the FP32 throughput of the RTX PRO 4000 Blackwell SFF (105 vs 24 TFLOPS dense). The GeForce RTX 5090 also carries 8 GB more memory (32 GB vs 24 GB).

RTX PRO 4000 Blackwell SFF: Blackwell, 2025 GeForce RTX 5090: Blackwell, 2025
FP32 dense lead
4.4x
GeForce RTX 5090: 105 vs 24 TFLOPS
VRAM
24 vs 32 GB
8 GB more for the GeForce RTX 5090
Bandwidth
432 GB/s vs 1.8 TB/s
GeForce RTX 5090 moves data faster
TDP
70 W vs 575 W
RTX PRO 4000 Blackwell SFF draws 505 W less

Specifications

RTX PRO 4000 Blackwell SFF
GeForce RTX 5090
Architecture
BlackwellBlackwell
Launch Year
20252025
Form Factor
PCIePCIe
VRAM
24 GB32 GB
Memory Bandwidth
432 GB/s1.8 TB/s
TDP
70 W575 W
Process Node
4nm4nm

Performance (TFLOPS)

RTX PRO 4000 Blackwell SFF
GeForce RTX 5090
FP64
No verified data available1.6 TFlops
FP32
24.0 TFlops 104.8 TFlops
TF32
No verified data available104.8 TFlops
209.5 TFLOPS sparse
BF16
No verified data available209.5 TFlops
419.0 TFLOPS sparse
FP16
No verified data available419.1 TFlops
sparse not published
FP8
No verified data available838.2 TFlops
sparse not published
FP6
No verified data available No verified data available
FP4
385.0 TFlops
770.0 TFLOPS sparse
1676.0 TFlops
sparse not published
INT8
No verified data available838.2 TFlops
sparse not published

FLOPS by Precision

What actually differs

Both GPUs launched in 2025: the RTX PRO 4000 Blackwell SFF on Blackwell and the GeForce RTX 5090 on Blackwell.

Dense throughput for the RTX PRO 4000 Blackwell SFF against the GeForce RTX 5090: FP32 24 vs 105 TFLOPS. Sparse figures, where the vendor publishes them, appear under each dense number in the performance table.

Memory is 24 GB against 32 GB, fed at 432 GB/s versus 1.8 TB/s. More memory per GPU means larger models fit before you have to shard across cards, which often matters more than raw TFLOPS for inference.

Power budgets are 70 W for the RTX PRO 4000 Blackwell SFF and 575 W for the GeForce RTX 5090. At FP32 that works out to 0.34 against 0.18 TFLOPS per watt.

For LLM training and inference, weight the FP8 and FP16 rows and memory capacity most heavily. For scientific and HPC workloads, the FP64 row is the one to read.

Frequently asked questions

Is the RTX PRO 4000 Blackwell SFF faster than the GeForce RTX 5090?

At FP32 precision the GeForce RTX 5090 reaches 105 TFLOPS dense against 24 TFLOPS for the RTX PRO 4000 Blackwell SFF. The performance table on this page lists every published precision for both GPUs.

Which has more memory, the RTX PRO 4000 Blackwell SFF or the GeForce RTX 5090?

The GeForce RTX 5090 carries 32 GB of VRAM versus 24 GB for the RTX PRO 4000 Blackwell SFF. Memory bandwidth is 432 GB/s for the RTX PRO 4000 Blackwell SFF and 1.8 TB/s for the GeForce RTX 5090.

How much power do the RTX PRO 4000 Blackwell SFF and the GeForce RTX 5090 draw?

The RTX PRO 4000 Blackwell SFF is rated at 70 W TDP and the GeForce RTX 5090 at 575 W. On FP32 throughput per watt, the RTX PRO 4000 Blackwell SFF is the more efficient part.

Can I rent the RTX PRO 4000 Blackwell SFF or the GeForce RTX 5090 in the cloud?

Yes. Live cloud listings tracked by Flopper start at $0.33 per GPU hour. The rental pricing section on this page lists current providers and rates for both GPUs.

Where to Rent

NVIDIA GeForce RTX 5090 32GB

ProviderConfigurationPrice/GPU-hrChecked
Vast.ai
2× GeForce RTX 5090 32GB Community $0.33 1h ago View →
Excess Supply
1× GeForce RTX 5090 32GB $0.65 1h ago View →
RunPod
1× GeForce RTX 5090 32GB Community $0.69 1h ago View →
Phyntec
8× GeForce RTX 5090 32GB $0.70 1h ago View →
RunPod
1× GeForce RTX 5090 32GB $0.99 1h ago View →
All GeForce RTX 5090 listings and price history →

Get Comparison Updates

New GPUs added weekly. Be the first to see how they compare.