GPU MODELS / RTX 3080

RTX 3080 cloud GPU pricing

19 listings across 10 providers. Cheapest published rate: $0.060/GPU/hr at Salad. Sorted by per-GPU hourly price. Every row links to the provider's own published price.

WHAT IS THE RTX 3080

The card that defined pandemic-era GPU scarcity remains a catalog staple: fast GDDR6X bandwidth, Ampere tensor cores, and a 10-12 GB ceiling. It serves 7B-class models and media workloads well at budget rates - a known quantity, priced like one.

MAKER
NVIDIA
ARCHITECTURE
Ampere
LAUNCHED
2020
MEMORY
10 or 12 GB GDDR6X
MEMORY BANDWIDTH
760 GB/s (912 GB/s for the 12 GB)
TENSOR COMPUTE
29.8 TFLOPS FP32; 3rd-gen tensor cores
POWER DRAW
320 W (350 W for the 12 GB)
INTERCONNECT
PCIe 4.0 x16

WHAT IT IS GOOD AT

7B-class inference
Good - 760 GB/s serves quantized 7B models comfortably.
Media generation
Proven diffusion workhorse.
Fine-tuning
LoRA on 7B-class models at most.

WHO SHOULD NOT RENT THE RTX 3080

13B+ models - even the 12 GB variant runs out.
FP8 stacks - Ampere has no hardware FP8.
ProviderConfigVRAMLocation$/hr$/GPU/hrSource
QuickPodQuickPodRTX 3080 × 110 GBNot published$0.030$0.030STALEreceipt
SaladSaladRTX 3080 × 110 GBGlobal (distributed)$0.060$0.060receipt
SpotGPUsSpotGPUsRTX 3080 × 110 GBNot published$0.060$0.060STALEreceipt
gpu.aigpu.aiRTX 3080 × 110 GBNot published$0.130$0.130receipt
Gitee AIGitee AIRTX 3080 × 110 GBChina$0.132$0.132receipt
FeaturizeFeaturizeRTX 3080 × 110.5 GBChina$0.139$0.139receipt
AutoDLAutoDLRTX 3080 × 110 GBChina$0.139$0.139receipt
Hostnot GPUHostnot GPURTX 3080 × 110 GBAsia-Pacific$0.150$0.150receipt
Hostnot GPUHostnot GPURTX 3080 × 110 GBUS West$0.150$0.150receipt
Hostnot GPUHostnot GPURTX 3080 × 210 GBUS East$0.320$0.160receipt
RunPodRunPodRTX 3080 × 110 GBGlobal$0.170$0.170receipt
Hostnot GPUHostnot GPURTX 3080 × 110 GBUS East$0.170$0.170receipt
Hostnot GPUHostnot GPURTX 3080 × 110 GBEU West$0.180$0.180receipt
Hostnot GPUHostnot GPURTX 3080 × 210 GBEU West$0.360$0.180receipt
Hostnot GPUHostnot GPURTX 3080 × 210 GBAsia-Pacific$0.530$0.265receipt
GPUBrazilGPUBrazilRTX 3080 × 110 GBAsia-Pacific$0.355$0.355receipt
GPUBrazilGPUBrazilRTX 3080 × 110 GBEurope$0.355$0.355receipt
GPUBrazilGPUBrazilRTX 3080 × 110 GBSouth America$0.355$0.355receipt
GPUBrazilGPUBrazilRTX 3080 × 110 GBNorth America$0.740$0.740receipt

THE RTX 3080 MARKET, BY THE NUMBERS

PROVIDERS
10
CONFIGURATIONS
19
MEDIAN $/GPU/HR
$0.17
SPREAD
12.3x

10 providers publish 19 RTX 3080 configurations on the panel today. Published per-GPU hourly rates run from $0.060 (Salad) to $0.74 (GPUBrazil), a 12x spread between the cheapest and the most expensive published rate for the same chip. The median listing sits at $0.17/GPU/hr. 11 rows were reported in stock by the provider at last observation; the rest are published catalog rates. All rates are on-demand - committed-use and reserved pricing is excluded.

CHEAPEST RTX 3080 BY GEOGRAPHY
chinaGitee AI$0.132/hrunited statesHostnot GPU$0.150/hr
RTX 3080 PRICE TREND, DAILY PANEL SNAPSHOTS

Across every RTX 3080 row on the panel, the daily floor held flat at $0.060/GPU/hr over 53 days of tracking. Snapshots run daily; the full history is public in the repo.

RTX 3080 PROVIDER NOTES
Salad · 1 config · 1 region
Sets the panel floor for this model.
from $0.060/hr
gpu.ai · 1 config
Published catalog rates.
from $0.130/hr
Gitee AI · 1 config · 1 region
Published catalog rates.
from $0.132/hr
Featurize · 1 config · 1 region
Published catalog rates.
from $0.139/hr
AutoDL · 1 config · 1 region
Published catalog rates.
from $0.139/hr
Hostnot GPU · 7 configs · 4 regions · 7 in stock
Widest regional footprint here - 4 regions.
from $0.150/hr
RunPod · 1 config · 1 region
Published catalog rates.
from $0.170/hr
GPUBrazil · 4 configs · 4 regions · 4 in stock
Widest regional footprint here - 4 regions.
from $0.355/hr
BUYING RTX 3080: WHAT THE PANEL SAYS

At the current floor, one RTX 3080 running around the clock costs about $44 a month (730 hours of on-demand arithmetic). Per-GPU floors by config size: 1x $0.06 · 2x $0.16. Compare the alternatives below before committing - the same budget often buys more than one chip class.

RTX 3080 VS THE ALTERNATIVES

RTX 3080 vs RTX 3080 Ti
The Ti adds 2 GB, 20% more bandwidth, and ~15% more compute for a small premium - worth it when the spread is narrow.
RTX 3080 vs RTX 4070
Same 12 GB class, one generation newer, lower power. Similar model fit - pick on panel price.

Compare live rates: RTX 3080 vs RTX 3080 Ti · RTX 3080 vs RTX 4070

RTX 3080 PRICING FAQ

What is the cheapest RTX 3080 cloud GPU right now?
The cheapest published RTX 3080 rate on the panel is $0.060 per GPU-hour at Salad (RTX 3080 x 1, Global (distributed)). Every price links to the provider's own published rate.
How much does a RTX 3080 cost per hour?
Across 10 providers tracked today, published per-GPU hourly rates for RTX 3080 run from $0.06 to $0.74, with a median of $0.17. Committed-use and reserved rates are excluded; everything shown is on-demand.
How many providers rent RTX 3080 GPUs?
10 providers publish 19 distinct RTX 3080 configurations on Compute Cafe. 11 of those rows were reported in stock by the provider's live endpoint at last observation.
Where is RTX 3080 cheapest?
By published per-GPU hourly rate, the cheapest geography for RTX 3080 today is china at $0.13/hr (Gitee AI). The spread across 2 tracked geographies runs to $0.15/hr in united states.
Are RTX 3080 prices going up or down?
Over the 53 days Compute Cafe has tracked the panel, the RTX 3080 floor is holding flat: $0.060 to $0.060/GPU/hr (+0.0%). Daily snapshots are public in the repo, so the trend is auditable.
What does a RTX 3080 cost per month?
At the current floor of $0.060/GPU/hr, a RTX 3080 running 24/7 costs about $44 per month (730 hours of on-demand arithmetic on the cheapest published rate). Committed-use contracts can price lower; the panel tracks on-demand rates only.
Is the RTX 3080 still worth renting?
For 7B-class serving and media work, yes - it is a known quantity at budget rates. For anything past that, 10-12 GB is the ceiling.

See also alternatives to the RTX 3080 · cheapest GPU-hour per model · inference chips · training chips · by location