Skip to content
Launch imminent

Official date to be announced shortlyGPU offers available at launch; figures shown are indicative.

Priority access: sign up

RTX PRO 6000 Blackwell vs RTX 5090

Manufacturer specifications of the RTX PRO 6000 Blackwell and the RTX 5090, side by side, per GPU. A value the manufacturer does not publish for the exact model is marked as such, never estimated.

NVIDIA RTX PRO 6000 Blackwell 96GB

Blackwell

GPU memory
96 GB
Memory bandwidth
1,792 GB/s

Live price on EverXCloud · from $1.69 per GPU-hour

NVIDIA GeForce RTX 5090 32GB

Blackwell

GPU memory
32 GB
Memory bandwidth
1,792 GB/s

Live price on EverXCloud · from $0.69 per GPU-hour

Key differences

  • The RTX PRO 6000 Blackwell has 3.0× the GPU memory of the RTX 5090 (96 GB vs 32 GB).
  • Both have about the same memory bandwidth (1,792 vs 1,792 GB/s).

Specifications

RTX PRO 6000 Blackwell and RTX 5090 specifications, per GPU
SpecificationRTX PRO 6000 BlackwellRTX 5090
GPU memory96 GB (higher)32 GB
Memory typeGDDR7GDDR7
Memory bandwidth1,792 GB/s1,792 GB/s
Max powerNot published575 W
ArchitectureBlackwellBlackwell
InterconnectPCIe 5.0PCIe 5.0

The higher value of each measured row is shown in bold. Manufacturer specifications, per GPU. Tensor throughput is shown without sparsity. Values the manufacturer does not publish for this exact model are left out.

What fits in memory, on one GPU

Cautious model-size estimates for one RTX PRO 6000 Blackwell and one RTX 5090
WorkloadRTX PRO 6000 BlackwellRTX 5090
LLM inference, 16-bit≈ 35B parameters≈ 10B parameters
LLM inference, 4-bit≈ 120B parameters≈ 40B parameters
Full training, mixed precision≈ 4B parameters≈ 1B parameters

Cautious estimates counting 80% of memory for model weights; context length, batch size and framework change the real limit. Several GPUs together hold larger models.

RTX PRO 6000 Blackwell vs RTX 5090: frequently asked questions

Which has more memory, the RTX PRO 6000 Blackwell or the RTX 5090?

The RTX PRO 6000 Blackwell has 96 GB of GPU memory, against 32 GB for the RTX 5090.

Which is faster for LLM inference, the RTX PRO 6000 Blackwell or the RTX 5090?

Their memory bandwidth is about the same, so token generation speed is close for a model that fits on both; memory size decides which models fit. Measure your own model before a large run.

Can I rent the RTX PRO 6000 Blackwell and the RTX 5090 by the hour?

Yes, when sellers list them on the EverXCloud spot exchange: you buy GPU hours at the live price from a prepaid balance in euros, US dollars or pounds. Prices are shown excluding tax and EverXCloud is the merchant of record for every rental.

Be ready on launch day

Create your account now: you can top up your balance and rent the GPU that fits your workload as soon as the market opens.

Have GPU servers?

Businesses can join the EverXCloud Network and sell hours of their GPU servers on the spot exchange.

Become a supplier

Other comparisons