RTX 5090 vs RTX 5080
The RTX 5080 rents for 22% less per hour today — $0.53 against $0.69 — with 16 GB against 32 GB of memory.
Prices collected 2026-09-18 · 5 providers price both
Index median per GPU-hour, on-demand, per snapshot.
Side by side
| RTX 5090 | RTX 5080 | |
|---|---|---|
| Median rental, per GPU-hour | $0.69 | $0.53 |
| Cheapest listing | $0.47 | $0.26 |
| Per month at 720 h | $497 | $385 |
| Providers pricing it | 9 | 5 |
| Memory | 32 GB GDDR7 | 16 GB GDDR7 |
| Memory bandwidth | 1,792 GB/s | 960 GB/s |
| GB of memory per $/hr | 46.4 | 29.9 |
| Architecture | Blackwell | Blackwell |
| Class | Consumer | Consumer |
| Released | 2025-01 | 2025-01 |
Same provider, both GPUs
The cleanest comparison: 5 providers price both cards, so the difference is the GPU and nothing else. The RTX 5090 is cheaper at 0 of them.
| Provider | RTX 5090 | RTX 5080 | RTX 5080 premium |
|---|---|---|---|
 SaladCloud | $0.50 | $0.26 | -48% |
 Vast.ai | $0.68 | $0.53 | -21% |
 RunPod Community Cloud | $0.69 | $0.39 | -43% |
 GPU.ai | $0.73 | $0.67 | -8% |
 RunPod Secure Cloud | $0.99 | $0.59 | -40% |
What each one runs, and what a million tokens costs
Weights + KV cache at 8K context + overhead against 90% of each card's memory. Where a model needs more than one card the cell says how many and prices the set at today's median. $/M tokens is the hourly median divided by estimated single-stream output, from memory bandwidth — an estimate for ranking, not a benchmark.
| Model | Bits | RTX 5090 | RTX 5080 |
|---|---|---|---|
| Kimi K3 | 4-bit | 55× GPU$37.95/hr | 110× GPU$58.84/hr |
| Kimi K3 | 16-bit | 195× GPU$134.55/hr | 390× GPU$208.61/hr |
| DeepSeek V4.1-Flash | 4-bit | 11× GPU$7.59/hr | 22× GPU$11.77/hr |
| DeepSeek V4.1-Flash | 16-bit | 39× GPU$26.91/hr | 77× GPU$41.19/hr |
| Llama 4 Maverick | 4-bit | 8× GPU$5.52/hr | 16× GPU$8.56/hr |
| Llama 4 Maverick | 16-bit | 28× GPU$19.32/hr | 56× GPU$29.95/hr |
| gpt-oss-120b | 4-bit | 3× GPU$2.07/hr | 5× GPU$2.67/hr |
| gpt-oss-120b | 16-bit | 9× GPU$6.21/hr | 17× GPU$9.09/hr |
| Llama 3.1 70B | 4-bit | 2× GPU$1.38/hr | 4× GPU$2.14/hr |
| Llama 3.1 70B | 16-bit | 6× GPU$4.14/hr | 11× GPU$5.88/hr |
| Gemma 4 31B | 4-bit | ✓ 21.2 GB~82 tok/s · $2.34/M | 2× GPU$1.07/hr |
| Gemma 4 31B | 16-bit | 3× GPU$2.07/hr | 5× GPU$2.67/hr |
| Qwen3 Coder 30B-A3B | 4-bit | ✓ 19 GB~772 tok/s · $0.25/M | 2× GPU$1.07/hr |
| Qwen3 Coder 30B-A3B | 16-bit | 3× GPU$2.07/hr | 5× GPU$2.67/hr |
| Qwen3.8 27B | 4-bit | ✓ 18.7 GB~94 tok/s · $2.04/M | 2× GPU$1.07/hr |
| Qwen3.8 27B | 16-bit | 2× GPU$1.38/hr | 4× GPU$2.14/hr |
| gpt-oss-20b | 4-bit | ✓ 13.7 GB~708 tok/s · $0.27/M | ✓ 13.7 GB~379 tok/s · $0.39/M |
| gpt-oss-20b | 16-bit | 2× GPU$1.38/hr | 4× GPU$2.14/hr |
| Llama 3.1 8B | 4-bit | ✓ 6.9 GB~319 tok/s · $0.6/M | ✓ 6.9 GB~171 tok/s · $0.87/M |
| Llama 3.1 8B | 16-bit | ✓ 18.4 GB~90 tok/s · $2.13/M | 2× GPU$1.07/hr |
Buy or rent?
Hours of renting at today's median that add up to the card's approximate list price. List prices are typical quoted figures, not tracked, and do not include the server, power or the person to run it — so the break-even is a floor, not a verdict.
Questions
Is the RTX 5090 or the RTX 5080 cheaper to rent?
On 2026-09-18 the RTX 5090 has a market median of $0.69 per GPU-hour across 9 providers and the RTX 5080 $0.53 across 5, so the RTX 5080 is 22% cheaper per hour. Among the 5 providers that price both, the RTX 5090 is cheaper at 0 and the RTX 5080 at 5. Per hour is only half the answer — the table below prices a whole model on each.
Which should I rent for a 70B model?
At 4-bit, Llama 3.1 70B needs about 44.5 GB. On the RTX 5090 that is 2 cards at $1.38/hr; on the RTX 5080 4 cards at $2.14/hr.
What is the real difference between them?
32 GB GDDR7 at 1,792 GB/s (Blackwell) against 16 GB GDDR7 at 960 GB/s (Blackwell). Memory decides what fits; bandwidth decides how fast it generates, because each token streams the active weights once. Compute matters for prompt processing and training, not for how fast text appears.