BTC
ETH
USDT
XMR
LTC
GPU comparison · Blackwell vs Blackwell
NVIDIA RTX 5090 vs RTX 5080.
Two Blackwell GeForce cards with GDDR7 and FP4 tensor cores. The RTX 5090 has twice the memory (32 GB against 16 GB), about 1.9× the bandwidth and 1.9× the dense FP8 and BF16 throughput; the RTX 5080 costs $130 less per card per month. Per GB of memory the RTX 5090 is the cheaper of the two.
- 32 vs 16 GBGPU memory
- 1.79 vs 0.96 TB/smemory bandwidth
- $329 vs $199per GPU, per month
RTX 5090 vs RTX 5080: which one to rent.
Choose the RTX 5090 if
- Your model needs more than 16 GB: 32 GB holds about 24B parameters in FP8 or 45B in 4-bit per card by our sizing rule.
- You generate video or high-resolution images in batches, where 16 GB runs out first.
- You want the lower price per GB of GPU memory: $10.28 against $12.44 a month.
Choose the RTX 5080 if
- Your model fits in 16 GB: up to about 12B parameters in FP8 or 22B in 4-bit per card by our sizing rule.
- You want the lowest price per card: $199 against $329 a month.
- You run many small jobs side by side: a server of four RTX 5080 costs $796 a month.
RTX 5090 vs RTX 5080 specs, side by side.
Figures from NVIDIA’s published specifications, per GPU. Tensor figures are peak dense throughput.
| Per GPU | NVIDIA RTX 5090 | NVIDIA RTX 5080 | RTX 5080 / RTX 5090 |
|---|---|---|---|
| Architecture | Blackwell | Blackwell | — |
| Form factor | PCIe Gen5 graphics card | PCIe Gen5 graphics card | — |
| GPU memory | 32 GB GDDR7, 512-bit | 16 GB GDDR7, 256-bit | 0.5× |
| Memory bandwidth | 1,792 GB/s | 960 GB/s | 0.54× |
| GPU-to-GPU link | None (PCIe Gen5) | None (PCIe Gen5) | — |
| FP4 tensor (dense) | 1.7 PFLOPS | 0.9 PFLOPS | 0.54× |
| FP8 tensor (dense) | 419 TFLOPS | 225 TFLOPS | 0.54× |
| FP16 / BF16 tensor (dense) | 209.5 TFLOPS | 112.6 TFLOPS | 0.54× |
| FP64 | — | — | — |
| Max power | 575 W | 360 W | — |
| GPUs per server | 1, 2, 4 | 1, 2, 4 | — |
| Per GPU in our servers | 16 vCPU · 64 GB RAM · 1 TB | 8 vCPU · 32 GB RAM · 0.5 TB | — |
Sources: NVIDIA GeForce RTX 5090 · RTX Blackwell whitepaper · NVIDIA GeForce RTX 5080 · RTX Blackwell whitepaper. Full sheets: RTX 5090 · RTX 5080
RTX 5090 vs RTX 5080 rental price.
At GPUCenter, the RTX 5090 rents for $329 and the RTX 5080 for $199 per GPU per month, 30% below the market median of public on-demand prices, paid in crypto with no KYC. Hourly figures are the monthly price divided by 730 hours.
| GPUCenter | NVIDIA RTX 5090 | NVIDIA RTX 5080 | RTX 5080 / RTX 5090 |
|---|---|---|---|
| Price per GPU, per month | $329 | $199 | 0.6× |
| Equivalent per GPU-hour | $0.45 | $0.27 | — |
| Per GB of GPU memory, per month | $10.28 | $12.44 | 1.21× |
| Per PFLOPS of dense FP8, per month | $785 | $884 | 1.13× |
| Largest server | 4× · $1,316/mo | 4× · $796/mo | — |
| Market median, per GPU | $481 | $299* | — |
| Below the median | −32% | −33% | — |
* Indicative market reference: no provider publishes an on-demand price for this GPU yet. Same price per GPU at every server size, no hourly metering. How we compare prices · Configure an RTX 5090 server · Configure a RTX 5080 server
Which models fit on each.
Smallest server that holds each model, by precision. Rule of thumb with headroom for the KV cache, 92% of GPU memory usable.
| Model (total parameters) | RTX 5090 · FP8 | RTX 5080 · FP8 | RTX 5090 · 4-bit | RTX 5080 · 4-bit |
|---|---|---|---|---|
| Qwen3.5-9B | 1× | 1× | 1× | 1× |
| Gemma 4 31B | 2× | 4× | 1× | 2× |
| Llama 3.3 70B | 4× | — | 2× | 4× |
| gpt-oss-120b | — | — | 4× | — |
| DeepSeek-V4-Flash | — | — | — | — |
| Qwen3.5-397B-A17B | — | — | — | — |
| DeepSeek-R1 (671B) | — | — | — | — |
| Kimi K2.6 (1T) | — | — | — | — |
FP8 ≈ 1.2 bytes and 4-bit ≈ 0.65 bytes per parameter. — = larger than the biggest server of that GPU. Size another model · How much VRAM does an LLM need?
RTX 5090 vs RTX 5080 FAQ.
More on each GPU: NVIDIA RTX 5090 · NVIDIA RTX 5080.
Is the RTX 5090 worth the extra cost over the RTX 5080?
For AI work, yes when your model does not fit in 16 GB or memory bandwidth limits you: the RTX 5090 has twice the memory and about 1.9× the bandwidth and dense FP8 throughput for $130 more per card per month. When everything fits in 16 GB, the RTX 5080 is the cheaper card.
Which one runs larger models?
The RTX 5090: 32 GB against 16 GB per card. In a server of four cards, that is 128 GB (about 118 GB usable) against 64 GB (about 59 GB usable).
Do both come in multi-card servers?
Yes, in servers of 1, 2 or 4 cards connected over PCIe. GeForce cards have no NVLink.
Do both support FP4?
Yes. Both are Blackwell GPUs with FP4 tensor cores as well as FP8 and BF16. The previous GeForce generation (RTX 4090) stops at FP8.
Other GPU comparisons.
- Blackwell vs Hopper RTX 5090 vs H100 32 vs 80 GB · 1.8 TB/s vs 3.35 TB/s $329 vs $1,779/mo per GPU
- Blackwell vs Ada Lovelace RTX 5090 vs RTX 4090 32 vs 24 GB · 1.8 TB/s vs 1 TB/s $329 vs $269/mo per GPU
- Blackwell vs Ada Lovelace RTX 5080 vs RTX 4090 16 vs 24 GB · 960 GB/s vs 1 TB/s $199 vs $269/mo per GPU
Rent the RTX 5090 or the RTX 5080.
Dedicated servers with 1 to 4 GPUs, one monthly price, paid in crypto. Online in under 10 minutes, no KYC.