NVIDIA RTX 5090 GPU: Specs, VRAM, Price & Benchmarks (2026)
The NVIDIA RTX 5090 (32GB GDDR7) launched in 2025. This guide covers its full specs, VRAM, SXM/PCIe differences where they apply, AI performance, and live per-GPU rental pricing on Aquanode.
Short answer on cost: the RTX 5090 rents from $0.613 per GPU per hour on Aquanode.
How much VRAM does the RTX 5090 have?
The RTX 5090 has 32GB GDDR7, with 1,792 GB/s of derived memory bandwidth (bus width × transfer rate; not a vendor-stated figure).
- VRAM: RTX 5090 32GB GDDR7 vs RTX 4090 24GB GDDR6X
- VRAM: RTX 5090 32GB GDDR7 vs RTX 5080 16GB GDDR7
What fits in 32GB GDDR7 of VRAM
| Model | Precision | Fits? |
|---|---|---|
| Llama 3.1 8B | FP16 | ~16GB. Fits with solid headroom. |
| Mixtral 8x7B | INT4 | ~24GB. Fits on a single card. |
| Llama 3.1 70B | INT4 | ~35-40GB. Does not fit on a single 32GB card. |
| Mistral 7B | FP16 | ~14GB. Fits easily. |
Approximate, based on published parameter counts and standard bytes-per-parameter rules of thumb (FP16 ≈ 2 bytes/param, INT4 ≈ 0.5-0.6 bytes/param). Real footprint also depends on KV-cache size and framework overhead.
What can the RTX 5090 run?
Popular open models from small to frontier scale, with the memory each needs and how many RTX 5090 cards (32GB GDDR7 each) that takes.
| Model | As published | FP8 | INT4 |
|---|---|---|---|
| Qwen/Qwen3-8B 8.2B | BF16: ~18.3 GB, 1 GPU | FP8: ~9.2 GB, 1 GPU | INT4: ~4.6 GB, 1 GPU |
| Qwen/Qwen2.5-14B-Instruct 14.8B | BF16: ~33 GB, 2 GPUs | FP8: ~16.5 GB, 1 GPU | INT4: ~8.3 GB, 1 GPU |
| Qwen/Qwen3-32B 32.8B | BF16: ~73.2 GB, 3 GPUs | FP8: ~36.6 GB, 2 GPUs | INT4: ~18.3 GB, 1 GPU |
| Qwen/Qwen-72B 72.3B | BF16: ~162 GB, 6 GPUs | FP8: ~80.8 GB, 3 GPUs | INT4: ~40.4 GB, 2 GPUs |
| MiniMaxAI/MiniMax-M2.7 228.7B | FP8: ~256 GB, 8 GPUs | – | INT4: ~128 GB, 4 GPUs |
| deepseek-ai/DeepSeek-R1 684.5B | FP8: ~765 GB, 24 GPUs | – | INT4: ~383 GB, 12 GPUs |
Estimates: weights at the stated precision plus a flat 20% for KV cache and overhead, at a moderate context length. A dash means the precision is not offered for that model (it is already published at that size). INT4 needs a published quantized checkpoint. Open any model for a per-GPU breakdown, or use the RTX 5090 VRAM calculator.
RTX 5090 VRAM calculator: check which models fit in its memory at each precision.
All models that fit in 32 GB: the open models whose weights and overhead fit, at native, FP8 and INT4 precision.
RTX 5090 specs
| Architecture | NVIDIA, launched 2025 |
| VRAM | 32GB GDDR7 |
| Memory bandwidth | 1,792 GB/s (derived) |
| TDP | 575W |
| Form factor | PCIe, 2-slot |
Specs from the vendor's public product page. Memory bandwidth is marked (derived) because NVIDIA does not publish a bandwidth figure for this part. It is bus width × transfer rate, not a vendor-stated number. See the source.
Related reading: RTX 5090 for AI, RTX 5090 vs RTX 4090, RTX 5080 vs RTX 5090, and RTX PRO 6000 vs RTX 5090.
GPU Glossary: What is VRAM?, Tensor Cores, CUDA Cores, TFLOPS
RTX 5090 AI performance
The newest and highest-VRAM consumer flagship on this marketplace (32GB, Blackwell generation). Enough to fine-tune or serve larger 13B-30B-class models than a 24GB card can hold, at a still-below-data-center hourly rate.
- Memory bandwidth: 1,792 GB/s (derived)
Same limits as every consumer card here: no ECC, no NVLink, and a 32GB ceiling that rules out 70B-class models even quantized. High TDP (575W) also means fewer cards per rack than the workstation/data-center parts.
The cheapest RTX 5090 offer right now ($0.613/GPU/hr) is about 44% below the market median of $1.10/GPU/hr.
RTX 5090 price: what does it cost?
Buying. $1,999 (launch MSRP). Street prices have run well above MSRP since launch, commonly $2,500-$4,000+, due to limited supply and a broader GDDR7/VRAM shortage affecting the whole RTX 50 series.
Renting. The cheapest current on-demand rate for the RTX 5090 on Aquanode is $1.09/GPU/hr. Live rates range from $0.613 to $1.49 per GPU per hour, with a median of $1.10/GPU/hr. Billed by the offer's own terms; the table below shows every live rate.
| Region | $/GPU/hr | Available | VRAM | vCPU | RAM |
|---|---|---|---|---|---|
| Hong Kong | $0.613 | 4 | 32 GB | 32 | 63 GB |
| Sfo, Us | $0.658 | 1 | 32 GB | 102 | 217 GB |
| United States | $0.902 | 1 | 32 GB | 24 | 96 GB |
RTX 5090 price history
Aquanode stores one snapshot of its GPU price index per UTC day. For the RTX 5090 that is 25 days so far, 2026-10-06 to 2026-10-30, so this is a short history, not a long-run trend. The lowest per-GPU rate went from $0.587 on 2026-10-06 to $0.658 on 2026-10-30.
| Day (UTC) | Lowest per GPU hour | Median per GPU hour | Data-center lowest per GPU hour | Offers |
|---|---|---|---|---|
| 2026-10-30 | $0.658 | $1.09 | $1.09 | 90 |
| 2026-10-29 | $0.658 | $1.10 | $1.09 | 82 |
| 2026-10-28 | $0.658 | $1.16 | $1.09 | 67 |
| 2026-10-27 | $0.684 | $1.10 | $1.09 | 79 |
| 2026-10-26 | $0.658 | $1.10 | $1.09 | 69 |
| 2026-10-25 | $0.658 | $1.11 | $1.09 | 70 |
| 2026-10-24 | $0.658 | $1.10 | $1.09 | 46 |
| 2026-10-23 | $0.787 | $1.10 | $1.09 | 40 |
| 2026-10-22 | $0.658 | $1.10 | $1.09 | 51 |
| 2026-10-21 | $0.658 | $1.10 | $1.09 | 56 |
| 2026-10-20 | $0.658 | $1.10 | $1.09 | 61 |
| 2026-10-19 | $0.635 | $1.19 | $1.09 | 76 |
| 2026-10-18 | $0.658 | $1.20 | $1.09 | 71 |
| 2026-10-17 | $0.658 | $1.18 | $1.09 | 59 |
| 2026-10-16 | $0.614 | $1.09 | $1.09 | 53 |
| 2026-10-15 | $0.658 | $1.09 | $1.09 | 49 |
| 2026-10-14 | $0.627 | $1.09 | $1.09 | 62 |
| 2026-10-13 | $0.627 | $1.08 | $1.09 | 91 |
| 2026-10-12 | $0.558 | $0.882 | $1.09 | 82 |
| 2026-10-11 | $0.565 | $0.991 | $1.09 | 107 |
| 2026-10-10 | $0.569 | $0.973 | $0.903 | 84 |
| 2026-10-09 | $0.616 | $0.905 | $0.903 | 94 |
| 2026-10-08 | $0.543 | $0.909 | $0.903 | 113 |
| 2026-10-07 | $0.569 | $0.909 | $0.903 | 143 |
| 2026-10-06 | $0.587 | $0.905 | $0.903 | 99 |
Each row is the stored daily snapshot of the live index, copied as recorded. A dash means that day stored no figure. The same series is available as JSON and summarised in the monthly GPU price report.
How this price is calculated
All prices on this page are normalized to a per-GPU hourly rate using each offer's authoritative GPU count, so that raw price is divided by the number of GPUs it actually covers; some offers report price as already per-GPU, so those are used as-listed. An offer with a missing, zero, or invalid GPU count is excluded entirely rather than published at a guessed rate.
No offers were excluded from this snapshot for a missing or invalid price. No offers were dropped as price outliers in this snapshot.
Only the cheapest qualifying offer per provider is shown in the table above. This page regenerates at most once per hour.
RTX 5090 vs RTX 4090: how do they compare?
- VRAM: RTX 5090 32GB GDDR7 vs RTX 4090 24GB GDDR6X
- Memory bandwidth: 1,792 GB/s vs 1,008 GB/s
On Aquanode right now, RTX 4090 starts at $0.393/GPU/hr against RTX 5090's $0.613/GPU/hr, about 36% less.
RTX 5090 vs RTX 5080: how do they compare?
- VRAM: RTX 5090 32GB GDDR7 vs RTX 5080 16GB GDDR7
- Memory bandwidth: 1,792 GB/s vs 960 GB/s
On Aquanode right now, RTX 5080 starts at $0.478/GPU/hr against RTX 5090's $0.613/GPU/hr, about 22% less.
Compare the RTX 5090 with other GPUs
Compare RTX 5090 with
- B200 vs RTX 5090
- H200 vs RTX 5090
- H100 vs RTX 5090
- A100 vs RTX 5090
- DGX A100 vs RTX 5090
- RTX 5090 vs V100
- AMD MI300X vs RTX 5090
- L40S vs RTX 5090
- L40 vs RTX 5090
- A40 vs RTX 5090
- RTX 5090 vs RTX A6000
- RTX 5090 vs RTX A5000
- RTX 5090 vs RTX A4000
- L4 vs RTX 5090
- RTX 5090 vs T4
- RTX 5090 vs RTX 6000 Ada
- RTX 5090 vs RTX PRO 6000
- RTX 5090 vs RTX PRO 6000 WS
- RTX 5090 vs RTX PRO 6000 SE
- RTX 5090 vs RTX PRO 5000
- RTX 5080 vs RTX 5090
- RTX 5070 Ti vs RTX 5090
- RTX 5070 vs RTX 5090
- RTX 5060 Ti vs RTX 5090
- RTX 4090 vs RTX 5090
- RTX 4080 Super vs RTX 5090
- RTX 4080 vs RTX 5090
- RTX 4070 Ti vs RTX 5090
- RTX 4070 Super vs RTX 5090
- RTX 4070 vs RTX 5090
- RTX 4060 Ti vs RTX 5090
- RTX 3090 vs RTX 5090
- RTX 3080 vs RTX 5090
- RTX 3070 vs RTX 5090
- RTX 3060 vs RTX 5090
- B300 vs RTX 5090
- AMD MI355X vs RTX 5090
- AMD MI325X vs RTX 5090
- H100 NVL vs RTX 5090
- RTX 5090 vs RTX PRO 4500
- RTX 5090 vs RTX PRO 4500 SE
- RTX 5090 vs RTX PRO 4000
- RTX 5090 vs RTX 5880 Ada
- RTX 5000 Ada vs RTX 5090
- RTX 4000 Ada vs RTX 5090
- RTX 4000 SFF Ada vs RTX 5090
- RTX 2000 Ada vs RTX 5090
- RTX 4070 Ti Super vs RTX 5090
- RTX 5090 vs RTX A4500
- RTX 2080 Ti vs RTX 5090
- RTX 3080 Ti vs RTX 5090
- RTX 3070 Ti vs RTX 5090
- RTX 3060 Ti vs RTX 5090
- GTX 1080 Ti vs RTX 5090
- RTX 4060 vs RTX 5090
- RTX 5060 vs RTX 5090
- GTX 1660 Super vs RTX 5090
- RTX 2060 vs RTX 5090
- Radeon RX 9070 XT vs RTX 5090
- Radeon RX 6700 XT vs RTX 5090
- RTX 5090 vs RTX 6000
- A16 vs RTX 5090
- P40 vs RTX 5090
- P4 vs RTX 5090
- Quadro P2000 vs RTX 5090
- Quadro M4000 vs RTX 5090
Get notified when the price drops
GPU supply moves hourly. Tell us what you're waiting for and we'll email you when a matching offer appears across any provider we track.
Rent a RTX 5090 on Aquanode
- On-demand instances from $0.613/GPU/hr, billed by the provider's own terms, with no hardware procurement or long-term commitment.
- 95 live offers across 3 regions today.
- Set a price/availability alert above to hear the moment a cheaper or newly-available RTX 5090 offer appears.
- Compare every RTX 5090 offer side by side, or browse the full multi-provider GPU marketplace.
Good for
The newest and highest-VRAM consumer flagship on this marketplace (32GB, Blackwell generation). Enough to fine-tune or serve larger 13B-30B-class models than a 24GB card can hold, at a still-below-data-center hourly rate.
Not good for
Same limits as every consumer card here: no ECC, no NVLink, and a 32GB ceiling that rules out 70B-class models even quantized. High TDP (575W) also means fewer cards per rack than the workstation/data-center parts.
RTX 5090 FAQs
How much VRAM does the RTX 5090 have?
The RTX 5090 has 32GB GDDR7, with 1,792 GB/s of derived (bus width × transfer rate, not vendor-stated) memory bandwidth.
What is the NVIDIA RTX 5090?
The NVIDIA RTX 5090 is a GPU released in 2025, with 32GB GDDR7 of memory and a 575W power envelope. See the full spec table above for interconnect, form factor and tensor-throughput details.
How much does it cost to rent a RTX 5090?
Live RTX 5090 rental prices currently range from $0.613 to $1.49 per GPU per hour, with a median of $1.10 per GPU per hour.
What's the cheapest RTX 5090 rate?
The lowest current RTX 5090 rate on Aquanode is $0.613 per GPU per hour in Hong Kong.
How does the RTX 5090 compare to the RTX 4090?
VRAM: RTX 5090 32GB GDDR7 vs RTX 4090 24GB GDDR6X Memory bandwidth: 1,792 GB/s vs 1,008 GB/s On Aquanode right now, RTX 4090 starts at $0.393/GPU/hr against RTX 5090's $0.613/GPU/hr, about 36% less. See the full RTX 5090 vs RTX 4090 comparison for a shared-provider price breakdown.
How does the RTX 5090 compare to the RTX 5080?
VRAM: RTX 5090 32GB GDDR7 vs RTX 5080 16GB GDDR7 Memory bandwidth: 1,792 GB/s vs 960 GB/s On Aquanode right now, RTX 5080 starts at $0.478/GPU/hr against RTX 5090's $0.613/GPU/hr, about 22% less. See the full RTX 5090 vs RTX 5080 comparison for a shared-provider price breakdown.
What is the RTX 5090 good for?
The newest and highest-VRAM consumer flagship on this marketplace (32GB, Blackwell generation). Enough to fine-tune or serve larger 13B-30B-class models than a 24GB card can hold, at a still-below-data-center hourly rate.
What are the RTX 5090's limitations?
Same limits as every consumer card here: no ECC, no NVLink, and a 32GB ceiling that rules out 70B-class models even quantized. High TDP (575W) also means fewer cards per rack than the workstation/data-center parts.
Is renting cheaper than buying?
Renting avoids the upfront hardware cost and lets you match spend to actual usage. A rented RTX 5090 at $0.613/hr only costs money while it's running, whereas buying ties up capital in hardware that keeps depreciating whether it's in use or not. Buying outright runs $1,999 (launch MSRP). Which is cheaper depends on how continuously you'd run it; short or bursty workloads usually favor renting.
How is the RTX 5090 price calculated?
All prices are normalized to a per-GPU hourly rate using each offer's authoritative GPU count, which the raw price is divided by; some offers report price as already per-GPU. Offers whose price can't be safely normalized, or whose rate is an extreme outlier against the rest of the market, are excluded.
RTX 5090 price by region
Related guides
Other models in the same generation, then the rest of the GPU index.
Related on Aquanode: RTX 5090 price in Vietnam, MI325X vs RTX 5090 specs, RTX 2000 Ada vs RTX 5090 specs, GPU requirements for FLUX.1-dev, and GPU requirements for Gemma 3 12B. Browse the full lists: Compare any two GPUs side by side and GPU prices by country.
