Ranking 01
Cheapest
Lowest true cost per GPU-hour, in stock or stock unknown, newest reading per provider
NVIDIA · Blackwell · Consumer · 44 providers list itNVIDIA RTX 5090 cloud prices
Rented today by 25 providers with a current on-demand offer, from $0.24 to $1.88 per GPU-hour, median $0.69. Every number below links to the page it was read from, with its read date.
Memory32GB GDDR7
Bandwidth1.79TB/s
FP16 tensor—TFLOPS
FP32104TFLOPS
Power575W
Specifications: manufacturer figures.
Cheapest on-demand · read 21:35 UTC
$0.24/ GPU·h
SaladTrust 671× RTX 5090 · 32 GB · SaladCloud community nodes (global)- Median
- $0.69 25 providers, once each
- 30 days
- +29.7%
- 90 days
- —
Prices by rental mode · On-demand · Spot · Reserved · ServerlessRTX 5090 prices today
Cheapest, median and spread among current offers. Medians count one floor per marketplace and need at least three offers.
Cheapest$0.24Salad
Median$0.69n = 58
Highest$1.88among current offers
Current offers1,00625 providers · last 45 days
Ever listed1,12838 providers on file
Where today's on-demand offers sitone dot per eligible offer (58) · the pool behind the tiles · read Nov 1, 2026 21:35 UTC
Provider offer51Marketplace floor7Median$0.69Middle half$0.70 – $1.20
58 eligible offers from 25 providers, binned every $0.05: from $0.24 (Salad) to $1.88 (VoltageGPU). 755 further marketplace listings sit above their marketplace's floor and are not drawn (one floor per marketplace, as in the tiles). Hover a dot for the provider, price, configuration, region and read date; click it to open the provider.
On-demand price history · 92 daysdaily median across all providers · 8 provider lines from their 30-day series · read Nov 1, 2026
Median · all providers$0.69Providers
Today $0.6930 days +29.7%92 days —Low $0.52 · Aug 1High $1.03 · Oct 23
Daily median on-demand price per GPU-hour across all providers, 92 days to Nov 1, 2026. Provider lines: the provider's cheapest current on-demand offer, its listed price over the last 30 days (read Nov 1, 2026). Drag to zoom, double-click to reset.
Today's podium · on-demand · readings of the last 45 days · reference Nov 1, 2026 21:35 UTCWho rents the RTX 5090 best today
Cheapest true cost, best throughput per dollar, highest Trust Score. One provider per step, ten places, every row dated.
Ranking 02
Best value
Most throughput per dollar for LLM inference · Trust ≥ 60 first, then provisional · estimated
Ranking 03
Most trusted
Highest Trust Score among providers with an eligible offer · unscored providers follow, cheapest first
58 eligible on-demand offers from 25 providers · median $0.69 (n = 58) · not ranked: 186 unavailable, 1 resold, 4 older readings · readings of the last 45 days · reference Nov 1, 2026 · 21:35 UTCRankings are never sponsored. How we rank →
All offers · 2,092 readings on file · 44 providersEvery RTX 5090 offer, with its receipt
Current offers ranked by true cost, each with its read date. Older readings stay listed below, in grey, and never enter a ranking.
| Provider | Configuration | Region | Stock | List | True cost | Read | |
|---|---|---|---|---|---|---|---|
 Clore.aiTrust 46 | 1× RTX 509024 vCPU · 64 GB RAM · 905.883 GB disk | ET | Unavailable | $0.22 | $0.22 | 21:08 UTCfresh | Receipt |
| SaladTrust 67marketplace floor | 1× RTX 5090 | SaladCloud community nodes (global) | In stock | $0.24 | $0.24 | 21:10 UTCfresh | Receipt |
| Clore.aiTrust 46 | 1× RTX 509024 vCPU · 32 GB RAM · 34.292 GB disk | US | Unavailable | $0.26 | $0.26 | 21:08 UTCfresh | Receipt |
| Clore.aiTrust 46 | 1× RTX 509028 vCPU · 64 GB RAM · 796.282 GB disk | RU | Unavailable | $0.33 | $0.33 | 21:08 UTCfresh | Receipt |
| Clore.aiTrust 46 | 2× RTX 509028 vCPU · 96 GB RAM · 1,684.168 GB disk | India | Unavailable | $0.33 | $0.33 | 21:08 UTCfresh | Receipt |
| Clore.aiTrust 46 | 1× RTX 509048 vCPU · 64 GB RAM · 20,128 GB disk | RU | Unavailable | $0.33 | $0.33 | 21:08 UTCfresh | Receipt |
| Clore.aiTrust 46marketplace floor | 1× RTX 509016 vCPU · 48 GB RAM · 940.362 GB disk | China | In stock | $0.33 | $0.33 | 21:08 UTCfresh | Receipt |
| Clore.aiTrust 46 | 1× RTX 509032 vCPU · 64 GB RAM · 8.644 GB disk | Poland | Unavailable | $0.33 | $0.33 | 21:08 UTCfresh | Receipt |
| Clore.aiTrust 46 | 1× RTX 509016 vCPU · 32 GB RAM · 253,512 GB disk | Poland | Unavailable | $0.33 | $0.33 | 21:08 UTCfresh | Receipt |
| Clore.aiTrust 46 | 1× RTX 509024 vCPU · 128 GB RAM · 983.012 GB disk | Germany | Unavailable | $0.33 | $0.33 | 21:08 UTCfresh | Receipt |
| Clore.aiTrust 46 | 1× RTX 509032 vCPU · 64 GB RAM · 22.051 GB disk | Poland | Unavailable | $0.33 | $0.33 | 21:08 UTCfresh | Receipt |
 HyperAI | 1× RTX 5090 | — | Stock unknown | $0.35 | $0.35 | 21:11 UTCfresh | Receipt |
| 1× RTX 509032 vCPU · 63 GB RAM · 76 GB disk | Korea | Stock unknown | $0.37 | $0.37 | 21:13 UTCfresh | Receipt | |
| Clore.aiTrust 46 | 1× RTX 509024 vCPU · 32 GB RAM · 799.857 GB disk | US | Unavailable | $0.38 | $0.38 | 21:08 UTCfresh | Receipt |
| Clore.aiTrust 46 | 1× RTX 509012 vCPU · 64 GB RAM · 838.857 GB disk | RU | In stock | $0.38 | $0.38 | 21:08 UTCfresh | Receipt |
| Clore.aiTrust 46 | 1× RTX 50906 vCPU · 55 GB RAM · 380.857 GB disk | UA | Unavailable | $0.38 | $0.38 | 21:08 UTCfresh | Receipt |
| Clore.aiTrust 46 | 1× RTX 509016 vCPU · 32 GB RAM · 1,494.89 GB disk | Canada | Unavailable | $0.40 | $0.40 | 21:08 UTCfresh | Receipt |
| Clore.aiTrust 46 | 1× RTX 509016 vCPU · 32 GB RAM · 753.198 GB disk | Korea | In stock | $0.40 | $0.40 | 21:08 UTCfresh | Receipt |
| Clore.aiTrust 46 | 1× RTX 509016 vCPU · 64 GB RAM · 772.997 GB disk | BD | Unavailable | $0.40 | $0.40 | 21:08 UTCfresh | Receipt |
 Vast.aiTrust 78marketplace floor | 4× RTX 509064 vCPU · 62.1 GB RAM · 3,892 GB disk | SK | In stock | $0.40 | $0.40 | 21:08 UTCfresh | Receipt |
| Vast.aiTrust 78 | 2× RTX 509032 vCPU · 31.1 GB RAM · 1,946 GB disk | SK | In stock | $0.40 | $0.40 | 21:08 UTCfresh | Receipt |
| Vast.aiTrust 78 | 1× RTX 509016 vCPU · 15.5 GB RAM · 973 GB disk | SK | In stock | $0.40 | $0.40 | 21:08 UTCfresh | Receipt |
| Clore.aiTrust 46 | 2× RTX 509016 vCPU · 128 GB RAM · 1,535.226 GB disk | UA | Unavailable | $0.41 | $0.41 | 21:08 UTCfresh | Receipt |
| Clore.aiTrust 46 | 1× RTX 509016 vCPU · 64 GB RAM · 285.888 GB disk | Canada | Unavailable | $0.41 | $0.41 | 21:08 UTCfresh | Receipt |
| Clore.aiTrust 46 | 1× RTX 509016 vCPU · 64 GB RAM · 1,432.085 GB disk | Canada | Unavailable | $0.41 | $0.41 | 21:08 UTCfresh | Receipt |
1,006 fresh offers · the 25 cheapest shown here, all of them in the comparison tool · 122 older readings of the same provider replaced by a newer one · prices per GPU-hour · true cost includes the add-ons the provider publishesOpen in the comparison tool →
Value · what the hour buysPrice per unit, and what fits
The same hour, divided by what the card gives you; then the smallest real configuration for the most downloaded open models.
| Unit | At the cheapest | At the median |
|---|---|---|
| per GB of VRAM · hour32 GB | $0.007 | $0.022 |
| per TB/s of bandwidth · hour1.79 TB/s | $0.131 | $0.385 |
| Model | Needs | Smallest config | $ / month |
|---|---|---|---|
| all MiniLM L6 v2Sentence Transformers | 1 GBFP8 | 1× RTX 5090at Salad | $172 |
| paraphrase multilingual MiniLM L12 v2Sentence Transformers | 1 GBFP8 | 1× RTX 5090at Salad | $172 |
| bge-m3BAAI | 1 GBFP8 | 1× RTX 5090at Salad | $172 |
| Qwen3 0.6BQwen | 1 GBFP8 | 1× RTX 5090at Salad | $172 |
| all-mpnet-base-v2Sentence Transformers | 1 GBFP8 | 1× RTX 5090at Salad | $172 |
| bge-reranker-v2-m3BAAI | 1 GBFP8 | 1× RTX 5090at Salad | $172 |
Monthly = 730 hours at the cheapest current on-demand price for that GPU count · models sized at FP8 when available, weights plus runtime overheadSize any model →
Buying guide · 31 sourcesRenting the RTX 5090
Written from the prices we read on providers' own pages in October 2026: the cheapest hours we found, who to pick for which job, and the caveats behind each figure.
Published benchmarks on file
| FLUX.1 dev, 1024², secondes par image (lower is better) | Secondes |
|---|---|
| RTX 3090 | 26 |
| RTX 4090 | 11.28 |
| RTX 5090 | 8.78 |
| Qwen3 on RTX 5090, tokens/s (one user) | tok/s |
|---|---|
| Qwen3 8B | 185.91 |
| Qwen3 14B | 123.79 |
| Qwen3 32B | 61.38 |
| vLLM, aggregate throughput (tokens/s) | tok/s | GPUBank price / h |
|---|---|---|
| RTX 4090 | 2259 | $0.23 |
| RTX 5090 | 4570 | $0.48 |
Caveats and sources
- 1GPUBank pricing · read Oct 15, 2026 — GPUBank rates, inclusions, billing and top-up bonuses
- 2NVIDIA GeForce RTX 5090 product page · read Oct 15, 2026 — CUDA cores, memory, bus, power
- 3NVIDIA RTX 50 Series announcement · read Oct 15, 2026 — Launch date, $1,999 price, 1,792 GB/s, 680 Tensor Cores
- 4NVIDIA GeForce software licence · read Oct 15, 2026 — Clause 2.8, no datacenter deployment
- 5NVIDIA developer forum, "Driver EULA update: No Datacenter Deployment" · read Oct 14, 2026 — The December 2017 wording
- 6SaladCloud pricing · read Oct 16, 2026 — RTX 5090 at four priority tiers, vCPU and RAM included
- 7Vast.ai, Rent RTX 5090 GPUs · read Oct 16, 2026 — $0.37 headline on 15 Oct; no current offers on 14 and 16 Oct
- 8Vast.ai documentation, Billing · read Oct 15, 2026 — Storage billed while stopped, bandwidth per byte
- 9Vast.ai documentation, Quickstart · read Oct 15, 2026 — $5 minimum deposit
- 10Clore.ai, Rent RTX 5090 · read Oct 16, 2026 — Headline price, 21 Sep 2026 snapshot, bare-metal quotes, fees, per-minute billing
- 11Runpod pricing (updated 8 October 2026) · read Oct 16, 2026 — Secure $1.19 and Community $0.69 rates, machine, storage, $1.58 serverless
- 12CloudRift, Rent RTX 5090 GPUs · read Oct 16, 2026 — On-demand and reserved rates, vLLM benchmark
- 13Beam pricing · read Oct 15, 2026 — RTX 5090 machine at $0.72/hr, serverless rate
- 14HyperAI pricing · read Oct 16, 2026 — RTX 5090 at $0.35/hr, queues at peak hours
- 15ComputePrices, RTX 5090 · read Oct 14, 2026 — Shadeform, Novita AI and Akash rates (aggregator)
- 16GetDeploying, RTX 5090 rental prices · read Oct 16, 2026 — $0.72 median across 16 priced providers; HyperAI, GPUhub, GPU.ai, Oblivus rows (aggregator)
- 17Lium, RTX 5090 · read Oct 14, 2026 — Marketplace range
- 18Spheron GPU rental · read Oct 14, 2026 — Listed rate, no capacity on the day
- 19LeaderGPU, 8x RTX 5090 server · read Oct 15, 2026 — Monthly price excluding VAT
- 20Thunder Compute, "Vast.ai GPU Pricing 2026" · read Oct 14, 2026 — Typical Vast storage rate (a competitor's estimate)
- 21ComfyUI discussion #9002, FLUX.1 dev FP8 benchmark · read Oct 15, 2026 — Seconds per image on RTX 5090, 4090, 3090
- 22Hardware Corner, RTX 5090 LLM benchmarks · read Oct 15, 2026 — llama.cpp tokens per second, Qwen3 8B, 14B, 32B
- 23Presenc, RTX 5090 local LLM benchmarks · read Oct 14, 2026 — Llama 3.3 70B on two cards (DatabaseMart run)
- 24BestValueGPU, RTX 5090 price history · read Oct 14, 2026 — New and used street prices on 13 Oct 2026
- 25GPU Price History, RTX 5090 (US) · read Oct 15, 2026 — Lowest new listing on 11 Oct 2026
- 26EIA, Electric Power Monthly, table 5.3 · read Oct 15, 2026 — US residential price, January to July 2026
- 27SaladCloud blog, Upcoming SaladCloud price changes: September 12, 2026 · read Oct 14, 2026 — High, medium and low tiers raised on 12 Sep 2026
- 28GPUPoet, RTX 5090 price in October 2026 · read Oct 14, 2026 — Average of the three cheapest daily listings
- 29NVIDIA blog, GeForce RTX 50 Series launch · read Oct 14, 2026 — FLUX.1 dev in just over five seconds on the RTX 5090 with FP4 (vendor claim)
- 30LeaderGPU, GPU VPS with RTX 5090 · read Oct 14, 2026 — Single card at €449 a month excluding VAT
- 31Eurostat, household electricity prices, second half of 2025 · read Oct 15, 2026 — EU household average
RTX 5090 · comparisons and collectionsCompare the RTX 5090 side by side
Questions about renting the RTX 5090
How much does the RTX 5090 cost per hour?
Today the RTX 5090 rents from $0.24 per GPU-hour on-demand, with a median of $0.69 across 25 providers, each counted once at its cheapest current offer (58 current offers in all, readings of the last 45 days). Prices depend on the provider, the number of GPUs in the machine, the region and whether you rent on-demand, spot or reserved; true cost adds the mandatory add-ons each provider publishes.
Which cloud providers offer the RTX 5090?
29 providers have a current RTX 5090 offer in stock. Cheapest first: Salad ($0.24), Clore Ai ($0.33), Hyperai ($0.35), Gcube ($0.37), Vast Ai ($0.40), Gpuhub ($0.46). Older readings from other providers stay listed on this page, in grey, and are never ranked.
How much memory does the RTX 5090 have?
32 GB of GDDR7 at 1.79 TB/s. At FP8 that fits all MiniLM L6 v2 (1 GB), paraphrase multilingual MiniLM L12 v2 (1 GB), bge-m3 (1 GB), Qwen3 0.6B (1 GB), all-mpnet-base-v2 (1 GB) on a single card; larger models need two to eight GPUs or a different card.
Where is the RTX 5090 cheapest right now?
The lowest true cost among eligible offers is $0.24 per GPU-hour at Salad for a 1-GPU machine in SaladCloud community nodes (global), read within the ranking window. Offers marked unavailable, older than 45 days, superseded by a newer reading, resold or outside the outlier band are listed but not ranked.
Is the RTX 5090 cheaper on spot or reserved?
Spot (interruptible) capacity starts at $0.10 per GPU-hour; reserved terms start at $0.38. Spot suits jobs that checkpoint; reserved suits machines that run all month.