--- title: Where to Rent an RTX PRO 6000 at the Cheapest Price (Oct 2026) url: https://gpubank.io/rent-rtx-pro-6000 description: RTX PRO 6000 rental prices checked on 16 October 2026: from $0.33 interruptible and $0.95 on a marketplace to $1.15 all-in at GPUBank. Editions compared. generated: 2026-11-02T00:26Z publisher: GPUBank (GPU Compute OÜ, Tallinn, Estonia) summary_for_ai: https://gpubank.io/llms.txt full_text_for_ai: https://gpubank.io/llms-full.txt --- > Markdown version of https://gpubank.io/rent-rtx-pro-6000, generated from the live page. GPUBank is a GPU cloud billed by the second at one all-in price per GPU (vCPU, RAM, NVMe, a public IPv4 and traffic included), in Paris, Frankfurt, Ashburn and Singapore. Prices are in US dollars, excluding VAT. Updated 16 Oct 2026 · 38 offers checked # Where to rent an RTX PRO 6000 at the cheapest price We read every RTX PRO 6000 offer we could find on the provider's own page, worked out what each hour really includes and which edition of the card it gives you, and priced three real monthly workloads against buying the card. **GPUBank Research** 18 min read · Prices read on 16 Oct 2026 **96 GB**GDDR7 ECC **1.8 TB/s**memory bandwidth **FP4**Blackwell tensor cores **300 to 600 W**three editions The short answer, 16 Oct 2026 1. $0.33 /hr **Cheapest hour you can get.** A Vast.ai interruptible bid on a single 300 W Max-Q card, with the disk billed on top at $0.65 per GB-month and the card lost whenever a higher bid arrives. 2. $0.95 /hr **Cheapest marketplace listing.** A Max-Q card with 8 vCPU and 96 GB of RAM on a Vast.ai host in British Columbia, on PCIe Gen 4, disk and traffic extra; the cheapest full-power 600 W Workstation card started at $1.00. 3. $1.09 /hr **Cheapest dedicated machine at list price.** Beam's RTX PRO 6000 machine from $1.09 with 16 vCPU, 128 GB of RAM, a 1 TB NVMe drive and no egress fees; Beam's plans start at $20 a month with $20 of credit. 4. $1.15 /hr **Cheapest machine with everything included.** GPUBank at $1.15, billed per second, with 16 vCPU, 188 GB of RAM, a 2 TB NVMe disk, an IPv4 address and all traffic; about $0.72 effective after the 60 percent bonus on a $1,000 top-up, $0.92 reserved and $0.69 interruptible. [RTX PRO 6000 $1.15 /hr all-in Launch an RTX PRO 6000 Spot $0.69 · reserved $0.92 Billed per second](https://console.gpubank.io/launch?gpu=rtx-pro-6000) The RTX PRO 6000 is the professional Blackwell card that puts 96 GB of memory on a single PCIe board, and in October 2026 the hour sold under that name runs from a 33-cent interruptible bid to $5.50 per GPU on Azure, with GetDeploying's median across 165 on-demand configurations at $2.21. The spread is not noise. Each price buys a different machine, a different meter and a different promise about interruption, and it also buys one of three editions of the card that do not run at the same speed: a 600 W Workstation card, a 300 W Max-Q, or a passively cooled Server Edition with less memory bandwidth. A page that lines up those hours without saying which is which compares things that cannot be compared, so every row below says what the hour includes, how it is billed, the smallest unit you can rent and, where the provider says so, which edition you get. We started from our own survey of 29 September 2026 and re-read every figure on 16 October 2026 on the provider's own page, or on Vast.ai's public offer search, which lists the live listings behind its marketplace. Figures we could only see on an aggregator are labelled as such with the aggregator named. More than one price in five had moved in two weeks: Runpod's Secure Cloud went from $2.09 to $2.49, Verda's on-demand hour from $1.96 to $2.30, Nebius switched its preemptible capacity to a floating spot price on 8 October, and Vast.ai's own pricing pages for all three editions showed no offers while its search returned 81 rentable listings. This page is published by GPUBank, a GPU cloud run by GPU Compute OÜ in Tallinn, so read our own entry with that in mind: it is measured on the same terms as every other offer, and the text says where a marketplace, a spot tier or another cloud beats it. 01 / Three editions, one name ## Which RTX PRO 6000 you actually rent Three cards share the RTX PRO 6000 name, the same chip and the same 96 GB, but not the same speed. The Workstation Edition runs at up to 600 W with 1,792 GB/s, the Max-Q keeps that bandwidth under a 300 W cap with a blower cooler made for four cards in one tower, and the Server Edition is passive, configurable from 400 to 600 W, and rated at 1,597 GB/s. Two limits follow. Bandwidth sets the ceiling for single-stream token generation, so the Server Edition is up to about 11 percent slower than the other two on a one-user chat with a large model. Power sets the speed of compute-bound work such as diffusion, rendering and batched prefill, so the Max-Q gives up the most there: Puget Systems measured it 5 to 14 percent slower than the 600 W card in most content-creation tests, and 14 percent behind on Unreal Engine and DaVinci Resolve GPU effects. Edition 1 of 3 ### Workstation Edition **600 W** board power **Memory** 96 GB GDDR7 ECC **Cooling** Double flow-through, open-air cooler **Price** **$16,000** NVIDIA US marketplace, August 2026 (launch listing $8,565) **Who rents it** Vast.ai marketplace, 32 rentable listings from $1.00 on 16 Oct 2026; Lium shows a reference price but no pod 1,792 GB/s and the full 600 W: the fastest of the three for both decoding and compute-bound work, but its cooler suits one or two cards per desktop rather than racks. Edition 2 of 3 ### Max-Q Workstation Edition **300 W** board power **Memory** 96 GB GDDR7 ECC **Cooling** Blower, exhausts through the rear bracket **Price** **$22,256** GPU Poet, October 2026 average of the cheapest eBay listings (estimated MSRP $10,999) **Who rents it** Vast.ai marketplace, 25 rentable listings from $0.95 on demand and $0.33 interruptible on 16 Oct 2026 Same 1,792 GB/s, so single-user token speed matches the 600 W card, but Puget measured it 5 to 14 percent slower on compute-bound work. Edition 3 of 3 ### Server Edition **400 to 600 W** configurable **Memory** 96 GB GDDR7 ECC **Cooling** Passive, cooled by the server chassis **Price** **$23,288** GPU Poet, October 2026 average of the cheapest eBay listings; NVIDIA publishes no MSRP **Who rents it** Named by AWS (G7e), CoreWeave, Verda, Hyperstack and Lium; Vast.ai lists 24 offers from $1.33 on 16 Oct 2026 1,597 GB/s, about 11 percent less bandwidth than the workstation cards; its speed on compute-bound work depends on the power limit the host sets. Power gauges on one scale, 0 to 600 W; the hatched part of the Server Edition arc is the range a host can set. Prices are purchase prices of one card. The power limit a host sets matters as much as the edition. HOSTKEY found its Server Edition running at about 300 W in its default mode and measured a Wan 2.2 14B generation 25 percent shorter, 30 minutes per clip instead of 40, after switching it to the 600 W mode, and the Workstation card can be configured down to 150 W. A cheap listing can therefore be a power-capped card, in which case each job costs more than the hourly rate suggests. Clouds that name the edition sell the Server Edition: AWS, CoreWeave, Verda, Hyperstack and Lium all say so. Vast.ai lists all three as separate filters, and on 16 October its cheapest on-demand Max-Q cost $0.95, its cheapest Workstation card $1.00 and its cheapest Server card $1.33. GPUBank lists 1.8 TB/s, which matches the Workstation and Max-Q datasheets, but does not state the edition or power limit. ### What fits in 96 GB The table of fits shows why so many renters pick this card despite the edition question. Weights alone take about 2 bytes per parameter in 16-bit, 1.06 in 8-bit and 0.58 in 4-bit, and a model fits comfortably when they leave room for activations and cache. A 70B model fits in 8-bit with a modest cache and in 4-bit with room to spare, gpt-oss-120b fits in its native 4-bit form, Qwen3 32B, FLUX.1 dev and Wan 2.2's two 14B experts fit at any precision, and Mixtral 8x22B fits only in 4-bit, with about 14 GB left for context. Weights of each model on one 96 GB card Fits, room left for cache Tight, 85 to 96 GB Does not fit on one card Model **16-bit** 2 bytes **8-bit** 1.06 bytes **4-bit** 0.58 bytes **Llama 3.3 70B** 70.6B parameters **Qwen3 32B** 32.8B parameters **gpt-oss-120b** 117B parameters **Mixtral 8x22B** 141B parameters **FLUX.1 dev** 12B parameters **Wan 2.2 A14B (two 14B experts)** 27B parameters Weights only, at about 2, 1.06 and 0.58 bytes per parameter; gpt-oss-120b's 4-bit column is its native MXFP4 checkpoint. A model fits comfortably below about 85 GB, leaving room for activations and cache. Hover or tap a cell to read it on the 96 GB gauge. 02 / The card ## What the RTX PRO 6000 is and who rents it NVIDIA's RTX PRO 6000 Blackwell carries 96 GB of GDDR7 with ECC, fifth-generation Tensor Cores that compute in FP4, and a PCIe Gen 5 host interface, and it has no NVLink bridge, so work spread over several cards talks over PCIe. It is the GB202 die of the GeForce RTX 5090 with more of it switched on: 24,064 CUDA cores, 752 Tensor Cores, 188 RT cores and a 512-bit memory bus, with three times the memory. Multi-Instance GPU can split one card into four 24 GB or two 48 GB slices, which is why Runpod also sells 48 GB and 24 GB slices at $1.29 and $0.69 an hour, and why a cheap listing is worth checking for being a whole card. It launched in 2025 in three editions built around the same chip and the same memory. The Workstation Edition has a double flow-through cooler rated at 600 W and 1,792 GB/s of memory bandwidth, the Max-Q Workstation Edition keeps the bandwidth but is capped at 300 W, and the Server Edition, passively cooled for data-centre chassis, is configurable from 400 to 600 W and rated by NVIDIA at 1,597 GB/s and 4 PFLOPS of FP4. That one-ninth difference in bandwidth and the factor of two in power are why the edition matters, a point the section at the end of this page works through. Being a professional card is what puts it in clouds at all. NVIDIA's GeForce licence rules out data-centre deployment of consumer cards such as the RTX 5090, while the RTX PRO line is built and licensed for it, which is why AWS sells it in its G7e instances, Google in G4, Oracle in an 8-GPU bare-metal shape and CoreWeave in 8-GPU nodes, alongside specialised clouds and marketplaces. The people renting it are mostly teams that want a 70B-class model or gpt-oss-120b on one card instead of two or four smaller ones, creators running Wan 2.2 video and large image models without offloading weights to system RAM, and fine-tuners who want 96 GB for LoRA or QLoRA on large models. The price range follows from that mix: the same card is sold as a shared slice, a marketplace listing, a bid, a reservation, a serverless worker and one-eighth of a hyperscaler server, and each of those carries a different price for reasons that have little to do with the silicon. 03 / The price ladder ## How to read the RTX PRO 6000 market A managed on-demand cloud sells a fixed machine, the GPU with its vCPUs, memory and local disk, at one hourly price, and the differences between clouds lie in what else that price covers. GPUBank includes the IPv4 address and all traffic and bills only a persistent disk on top, Beam includes 16 vCPU, 128 GB of RAM and a 1 TB NVMe drive, Hyperstack bills only the GPU line with CPU and RAM included up to its stated maxima, and Runpod adds container and volume disk charges. A cheaper managed hour can still produce a larger bill, so the cost section below prices whole months rather than hours. One offer does not fit any of the usual boxes: packet.ai's $0.66 Dynamic tier is a full card's memory on a card shared between tenants under scheduler isolation, a different product from a dedicated hour at the same price; its page includes up to 2 TB of local NVMe per node and charges no egress, but publishes no price for storage that outlives an instance. On a peer-to-peer marketplace independent hosts post their own GPU rate and add disk and bandwidth as separate lines, and the machine changes from one listing to the next. On Vast.ai on 16 October the cheapest listing was a 300 W Max-Q at $0.95 with 8 vCPU on a PCIe Gen 4 host. Spot or interruptible capacity is cheaper because it can be taken back: Vast.ai bids start at $0.33, GPUBank's interruptible tier is $0.69 with five minutes of notice, and Nebius floats from $0.79. Reserved capacity runs from week-long terms to three-year commitments and is often sold by the 2-GPU or 8-GPU server, serverless workers bill active seconds at two to three times a dedicated hour, and the hyperscalers price the card at $3.36 to $5.50 per GPU, usually as part of a larger machine. *Average of 29 clouds, 29 Sep $2.27* **Managed on demand** A fixed machine at one hourly price. *Hostinger**$0.65*** *packet.ai**$0.66*** *HyperAI**$0.80*** *Beam**$1.09*** *GPUhub**$1.10*** *GPUBank**$1.15*** *Nebius**$1.80*** *Hyperstack**$1.85*** *Massed Compute**$2.19*** *Verda**$2.30*** *Runpod**$2.49*** **Marketplace** Hosts set their own price; machines vary. *Clore.ai**$0.92*** *Vast.ai**$0.95*** *Vast.ai**$1.00*** *Vast.ai**$1.33*** *Lium**$1.60*** *Runpod**$1.69*** **Spot and interruptible** Cheaper hours that can be taken back. *Vast.ai**$0.33*** *Nova Cloud**$0.66*** *GPUBank**$0.69*** *Nebius**$0.79*** *Azure**$1.02*** *Verda**$1.15*** *CoreWeave**$1.20*** **Reserved** Committed for a term, paid either way. *HyperAI**$0.48*** *Leaseweb**$0.88*** *Database Mart**$0.90*** *GPUBank**$0.92*** *CloudRift**$1.10*** *Hyperstack**$1.30*** **Serverless** Billed by active seconds. *Verda**$2.53*** *Runpod**$3.49*** USD per GPU-hour, read on 16 Oct 2026. Hover a dot for the offer; crowded dots keep their label in the table below. The large clouds are listed separately. 04 / Provider by provider ## What each price buys, and what it leaves out. The captures below were taken on 16 Oct 2026 from each provider's own RTX PRO 6000 or pricing page, so every figure can be read next to the conditions printed beside it. gpubank.io/pricing **GPUBank** All-in $1.15/hr Managed cloud · 1 to 8 GPUs 16 vCPU 188 GB RAM 2 TB NVMe IPv4 and all traffic Persistent disk $0.03/GB-month $0.72 effective from a $1,000 top-up GPUBank sells the RTX PRO 6000 at $1.15 an hour with 16 vCPU, 188 GB of RAM, a 2 TB local NVMe disk, an IPv4 address and all traffic, billed per second from the moment the machine passes its health check, with no minimum. The same card costs $0.69 interruptible, with five minutes of notice and a stop rather than a deletion, and $0.92 reserved for a week or a month. A top-up bonus of up to 60 percent brings the effective on-demand hour to $0.72, but payment is in crypto only and cards are not accepted for now. The pricing page gives 1.8 TB/s but not the edition or power limit, which is worth asking before a long reservation. Checked 16 Oct 2026 vast.ai/pricing/gpu/RTX-PRO-6000-WS **Vast.ai** from $0.95/hr Marketplace · all three editions GPU rate set by the host vCPU and RAM vary by listing Disk $0.13 to $0.87/GB-month Traffic up to $0.04/GB Bids from $0.33 Vast.ai is the one marketplace that sells all three editions under separate names, and its public offer search returned 81 rentable RTX PRO 6000 listings on 16 October 2026 even though its own pricing pages read "No current offers". The cheapest on-demand listing was a 300 W Max-Q at $0.95 with 8 vCPU and 96 GB of RAM in British Columbia on PCIe Gen 4, the cheapest Workstation card $1.00 and the cheapest Server card $1.33 on hosts of two to eight GPUs. Interruptible bids started at $0.33 for one Max-Q. Each host adds its own disk rate, between $0.13 and $0.87 per GB-month on these listings, and a bandwidth charge, and storage keeps billing while an instance is stopped. Billing is per second. The price is the lowest on the market; the machine, the edition and the host's reliability are yours to check. Checked 16 Oct 2026 beam.cloud/pricing **Beam** $1.09/hr Managed cloud · single-GPU machines 16 vCPU 128 GB RAM 1 TB NVMe Plans from $20/month with $20 credit Beam's pricing page lists an RTX PRO 6000 machine from $1.09 an hour with 16 vCPU, 128 GB of RAM and a 1 TB NVMe drive, and its own pricing article of 29 September 2026 gives the same rate with vCPU, RAM and NVMe included. That is the cheapest dedicated full machine we could read on a provider's own page, six cents below GPUBank's list price, with 60 GB less system memory. The page does not say whether $1.09 is the exact rate or a starting price, nor which edition the PCIe card is. Beam's plans start with a $20 Developer plan that includes $20 of credit each month, and its Team plan costs $89 a month plus usage, so a light user should count the plan in the hour. Card payment is the norm, which makes Beam the simplest alternative for a buyer who cannot pay in crypto. Checked 16 Oct 2026 docs.nebius.com/compute/resources/pricing **Nebius** $1.80/hr Managed cloud · 1 or 8 GPUs 24 vCPU 218 GB RAM Preemptible from $0.79, floating Storage extra Nebius lists the RTX PRO 6000 at $1.80 per GPU-hour on demand in us-central1, unchanged by its October price increases on data-centre cards. Its preemptible price moved on 8 October 2026 from a fixed $0.95 to dynamic spot pricing that starts at $0.79, can rise or fall with available capacity, and is capped one cent below the on-demand rate, with the current and past prices shown in the console. GetDeploying lists the instances at 24 vCPU and 218 GB of RAM per GPU in 1-GPU and 8-GPU presets. Storage and network are billed separately. Nebius is a reasonable choice for a team that already runs there, and its spot floor sits close to GPUBank's $0.69 interruptible hour, but for a single card bought on demand it costs 65 cents an hour more than GPUBank's all-in price. Checked 16 Oct 2026 hyperstack.cloud/gpu-pricing **Hyperstack** $1.85/hr Managed cloud · Server Edition Up to 31 pCPU Up to 180 GB RAM Root and ephemeral NVMe Spot $1.48 Reserved from $1.30 Volumes $0.07/GB-month Hyperstack's pricing page lists the RTX PRO 6000 SE, the Server Edition, at $1.85 an hour on demand, $1.48 on spot and from $1.30 reserved, with up to 31 pCPU and 180 GB of RAM per GPU and only the GPU billed. Shared volumes cost $0.000096774 per GB-hour, about $0.07 per GB-month, so a 500 GB volume adds $35.32 to a month. GetDeploying saw both the on-demand and spot flavours out of stock on 15 October, which leaves the reserved price as the main reason to look here: from $1.30 it undercuts most European and US clouds, though it remains above GPUBank's $0.92 week or month and well above the aggregator-only reservations of HyperAI and Leaseweb. The Server Edition label also tells you exactly which card you get. Checked 16 Oct 2026 runpod.io/pricing **Runpod** $2.49/hr Managed cloud and community · per second 16 vCPU 188 GB RAM Community Cloud $1.69 Serverless $3.49 Volume disk $0.20/GB-month idle Runpod's pricing page lists the RTX Pro 6000 at $2.49 an hour in its Secure Cloud with 16 vCPU and 188 GB of RAM, up from $2.09 on 29 September, at $1.69 in its Community Cloud, where independent hosts supply the machines, and at $3.49 for a 96 GB serverless worker. Billing is per second. Disk is where Runpod gets expensive for this card: a container disk costs $0.10 per GB-month, a volume disk $0.10 while the pod runs and $0.20 while it is idle, and network storage $0.07 under 1 TB. Kept for a month with the pod mostly stopped, a 500 GB volume costs close to $96, more than six times GPUBank's $15. Runpod's case is its tooling and template ecosystem, and its Community tier is the cheaper entry point when stock is there. Checked 16 Oct 2026 ### Where GPUBank wins, and where it does not GPUBank rents the RTX PRO 6000 at $1.15 per GPU-hour on demand, $0.69 interruptible and $0.92 reserved for a week or a month, each with 16 vCPU, 188 GB of RAM, a 2 TB NVMe disk, one IPv4 address and all inbound and outbound traffic, in machines of 1, 2, 4 or 8 GPUs. Billing is per second with no minimum, the meter starts after the health check, a failed launch is not billed, and the only extra line is a persistent disk at $0.03 per GB-month. Payment is a prepaid balance topped up in crypto from $25, and the top-up bonus brings the on-demand hour to about $0.96 from $25, $0.85 from $100 and $0.72 from $1,000. It is listed in Paris, Ashburn and Singapore, and the median launch to a machine that has passed its health check takes 48 seconds. Where it wins is the all-in month: no egress, no IP line and a cheap stopped disk made it the cheapest dedicated option in all three scenarios below, beaten on paper only by a shared card or a floating spot floor before their storage is counted, or by a marketplace bid whose disk is deleted between runs. Where it loses is the headline hour. Vast.ai bids at $0.33 and Max-Q listings at $0.95, packet.ai's shared $0.66, HyperAI's $0.80 and Beam's $1.09 dedicated machine are all below its $1.15 list price, and its interruptible tier at $0.69 is twice the cheapest Vast.ai bid. Its reservation at $0.92 needs only a week but is prepaid and cannot be refunded once it starts. It takes no cards for now, which rules it out for a buyer who can only pay by card. And its pricing page lists 96 GB GDDR7 ECC at 1.8 TB/s without naming the edition or power limit behind the rental, so a buyer whose job depends on sustained power should ask before reserving. ### The rest of the market Several offers below GPUBank's $1.15 list price could only be partly checked. packet.ai's $0.66 is on its own page but sits on a shared card, with the dedicated tier listed as launching soon. HyperAI's pricing page lists the card at $0.80 an hour without stating the vCPU, RAM or disk that come with it, and GetDeploying shows a one-month reservation at $0.48 that we could not find on HyperAI's site. Hostinger at $0.65 and GPUhub at $1.10 appear only on GetDeploying, which saw GPUhub in stock with 22 vCPU and 110 GB of RAM on 15 October. Clore.ai's own documentation carries a dated snapshot of 24 August 2026 that put the card at $0.92 to $1.38 with a median of $1.21, from only four free servers, and no live price could be read since. Lium's Server Edition pods were asking $1.60 to $1.99 per GPU on 8-GPU machines, billed per second. Above GPUBank's list price a group of European and US clouds sells dedicated single cards between $1.74 and $2.41: Nova Cloud at $1.74, UpCloud at $1.88, EcoHash at $1.89, Cloudzy at $1.93, Lyceum at $1.99, JarvisLabs at $2.19, Koyeb at $2.20 and Sesterce at $2.41, all from GetDeploying's crawl of 15 October, and Verda at $2.30 on its own page with 30 CPU threads and 90 GB of RAM per GPU. Massed Compute's own page lists the card at $2.19 with 16 vCPU, 144 GB of RAM and 725 GB of storage, and Exoscale sells it at $2.00 per GPU only as an eight-card node. Their case rests on something other than price: a specific European jurisdiction, card payment, or an existing account. On the spot side, Verda's interruptible instances cost $1.15 and its serverless containers $2.53, or $1.27 on spot, while CoreWeave sells 8-GPU Server Edition spot nodes at $9.56 an hour, $1.195 per GPU. CloudRift's own page quotes the card only reserved, $1.16 for a month and $1.10 for three, with on-demand capacity on request. The long-term reservations seen on aggregators, Leaseweb at $0.88 per GPU on a 2-GPU server for a month, Database Mart at $0.90 on a 24-month term and Utho at $1.18 on 36 months, need a quote before anyone plans around them. ### The large clouds, per GPU CoreWeave $2.50 /GPU-hr 8-GPU node · 128 vCPU, 1,024 GB RAM, 7.68 TB AWS $3.36 /GPU-hr 1 GPU · g7e.2xlarge, 8 vCPU, 64 GiB; EBS, egress extra Google Cloud $4.50 /GPU-hr 1 GPU · g4-standard-48, 48 vCPU, 180 GB Oracle Cloud $4.50 /GPU-hr 8-GPU node · Bare metal, 144 OCPU, 3 TB RAM Azure $5.50 /GPU-hr 1 GPU · 128 vCPU, 256 GB RAM AWS and CoreWeave name the Server Edition in their offers, and the large clouds cost two to five times a specialised cloud at list price. AWS sells it in G7e instances from the g7e.2xlarge, one GPU with 8 vCPU and 64 GiB, at $3.363 an hour on demand and about $2.56 on spot, up to the 8-GPU g7e.48xlarge with 192 vCPU and 2 TiB at $33.14, or $4.14 per GPU, with EBS storage and egress billed on top. Google's g4-standard-48 gives one GPU with 48 vCPU and 180 GB for $4.50 an hour in us-central1, $3.105 on a one-year commitment, $1.98 on three years and about $1.71 on spot. GetDeploying lists Azure's full-GPU size at $5.50 per GPU and Oracle's BM.GPU.RTXPRO.8 at $4.50 per GPU, the Oracle shape being a bare-metal server with 8 cards, 144 OCPU and 3 TB of RAM. CoreWeave's 8-GPU Server Edition node with 128 vCPU, 1,024 GB of RAM and 7.68 TB of local storage costs $20.00 an hour on its own page, which is $2.50 per GPU, the cheapest of this group but only by the whole node. These clouds make sense when the rest of a stack already runs there, when enterprise discounts bring the effective rate down, or when compliance demands a particular provider. For a team that only needs the card, the hyperscaler premium buys nothing the card itself does faster, and the per-GPU price of an 8-GPU instance is only reachable by renting all eight. 05 / Every offer ## The whole comparison, sorted by kind and price. | Provider | USD / GPU-hr | Kind | Unit | What the price includes | Billing | Stock | Basis | | --- | --- | --- | --- | --- | --- | --- | --- | | HHostinger | $0.65 | Managed on demand | 1 GPU | Inclusions not stated | Hourly | Not stated | Aggregator | | ppacket.ai | $0.66 | Managed on demand | 1 GPU | Shared card (multi-tenant), no egress fees | Hourly | In stock | Provider page | | HHyperAI | $0.80 | Managed on demand | 1 GPU | Machine not stated on the pricing page | Not stated | Not stated | Provider page | | Beam | $1.09 | Managed on demand | 1 GPU | 16 vCPU, 128 GB RAM, 1 TB NVMe | Not stated | Not stated | Provider page | | GGPUhub | $1.10 | Managed on demand | 1 GPU | 22 vCPU, 110 GB RAM | Not stated | In stock | Aggregator | | GPUBank | $1.15 | Managed on demand | 1 GPU | 16 vCPU, 188 GB RAM, 2 TB NVMe, IPv4, all traffic | Per second | In stock | Provider page | | Nebius | $1.80 | Managed on demand | 1 GPU | 24 vCPU, 218 GB RAM; storage extra | Per hour | Not stated | Provider page | | Hyperstack | $1.85 | Managed on demand | 1 GPU | Up to 31 pCPU, 180 GB RAM; Server Edition | Per minute | Sold out | Provider page | | Massed Compute | $2.19 | Managed on demand | 1 GPU | 16 vCPU, 144 GB RAM, 725 GB storage | Not stated | Not stated | Provider page | | Verda | $2.30 | Managed on demand | 1 GPU | 30 CPU threads, 90 GB RAM; Server Edition | Not stated | In stock | Provider page | | Runpod | $2.49 | Managed on demand | 1 GPU | 16 vCPU, 188 GB RAM; disks extra | Per second | In stock | Provider page | | Clore.ai | $0.92 | Marketplace | 1 GPU | Range $0.92 to $1.38, median $1.21 (24 Aug 2026) | Per minute | Not stated | Provider snapshot | | Vast.ai | $0.95 | Marketplace | 1 GPU | Max-Q 300 W, 8 vCPU, 96 GB RAM; disk $0.13/GB-month | Per second | In stock | Provider page | | Vast.ai | $1.00 | Marketplace | 1 GPU | Workstation 600 W, 16 vCPU; disk $0.20/GB-month | Per second | In stock | Provider page | | Vast.ai | $1.33 | Marketplace | 1 GPU | Server Edition on 2 to 8-GPU hosts | Per second | In stock | Provider page | | Lium | $1.60 | Marketplace | 8-GPU node | Server Edition pods, asks $1.60 to $1.99 | Per second | In stock | Provider page | | Runpod | $1.69 | Marketplace | 1 GPU | Community Cloud, 16 vCPU, 188 GB RAM | Per second | In stock | Provider page | | Vast.ai | $0.33 | Spot and interruptible | 1 GPU | Minimum bid, Max-Q; disk $0.65/GB-month | Per second | In stock | Provider page | | NNova Cloud | $0.66 | Spot and interruptible | 1 GPU | 22 vCPU, 82 GB RAM | Not stated | In stock | Aggregator | | GPUBank | $0.69 | Spot and interruptible | 1 GPU | As on demand, 5 min notice, stop not delete | Per second | In stock | Provider page | | Nebius | $0.79 | Spot and interruptible | 1 GPU | Floating price, floor shown | Per hour | Not stated | Provider page | | Azure | $1.02 | Spot and interruptible | 1 GPU | 128 vCPU, 256 GB RAM | Per second | Not stated | Aggregator | | Verda | $1.15 | Spot and interruptible | 1 GPU | 30 CPU threads, 90 GB RAM | Not stated | In stock | Provider page | | CoreWeave | $1.20 | Spot and interruptible | 8-GPU node | 128 vCPU, 512 GB RAM, 7.68 TB | Per hour | Not stated | Provider page | | HHyperAI | $0.48 | Reserved | 1 GPU | One month, 32 vCPU, 90 GB RAM | Per term | In stock | Aggregator | | LLeaseweb | $0.88 | Reserved | 2-GPU server | 2-GPU dedicated server, one month | Per term | Not stated | Aggregator | | DDatabase Mart | $0.90 | Reserved | 1 GPU | 24-month term, 32 vCPU, 84 GB RAM | Per term | Not stated | Aggregator | | GPUBank | $0.92 | Reserved | 1 GPU | Week or month, prepaid | Per term | In stock | Provider page | | CloudRift | $1.10 | Reserved | 1 GPU | Three months | Per term | Not stated | Provider page | | CloudRift | $1.16 | Reserved | 1 GPU | One month; on demand on request | Per term | Not stated | Provider page | | Hyperstack | $1.30 | Reserved | 1 GPU | Starting price, terms on request | Per term | Not stated | Provider page | | CoreWeave | $2.50 | Large clouds | 8-GPU node | 128 vCPU, 1,024 GB RAM, 7.68 TB | Per hour | Not stated | Provider page | | AWS | $3.36 | Large clouds | 1 GPU | g7e.2xlarge, 8 vCPU, 64 GiB; EBS, egress extra | Per second | Not stated | Aggregator | | Google Cloud | $4.50 | Large clouds | 1 GPU | g4-standard-48, 48 vCPU, 180 GB | Per second | Not stated | Aggregator | | Oracle Cloud | $4.50 | Large clouds | 8-GPU node | Bare metal, 144 OCPU, 3 TB RAM | Per hour | Not stated | Aggregator | | Azure | $5.50 | Large clouds | 1 GPU | 128 vCPU, 256 GB RAM | Per second | Not stated | Aggregator | | Verda | $2.53 | Serverless | 1 GPU | Containers, $1.27 on spot | Per second | Not stated | Provider page | | Runpod | $3.49 | Serverless | 1 GPU | 96 GB worker; storage extra | Per second | Not stated | Provider page | "Aggregator" rows could not be read on the provider's own page that day; "Provider snapshot" rows come from a page dated by the provider. 06 / The real monthly bill ## The real cost beyond the hourly rate Storage is the line that most often overturns an hourly comparison, because model weights for this card are large and live on disk between sessions. A 500 GB persistent disk costs $15.00 a month on GPUBank and nothing else is billed while the machine is stopped. The same 500 GB costs about $35.32 a month on Hyperstack at $0.000096774 per GB-hour, and on Runpod a volume disk costs $0.10 per GB-month while the pod runs and $0.20 while it is idle, so a mostly idle 500 GB volume approaches $100 a month. Vast.ai hosts set their own rate, from $0.13 to $0.87 per GB-month on the RTX PRO 6000 listings we read, and storage keeps billing while an instance is stopped. Traffic is the second line: AWS bills egress at about $0.09 per GB, so a video service sending 5 TB a month would pay roughly $441 for it after the free first 100 GB, more than half the card's own month on GPUBank. Google bills it too and Vast.ai hosts charge up to $0.04 per GB, while GPUBank, packet.ai and CoreWeave state none. Billing granularity decides the price of short work. A 61-minute job, 3,660 seconds at $0.000319, costs $1.17 on GPUBank, while the same rate billed per started hour costs $2.30 because the sixty-first minute buys a second hour. packet.ai bills hourly, Runpod and Vast.ai per second, and the hyperscalers per second after a one-minute minimum. Deposits and credit matter too: GPUBank's minimum top-up is $25 and its balance never expires, Beam's cheapest plan is $20 a month with $20 of credit, and a top-up bonus is a real discount only on money you will spend. The last cost is lost time. GPUBank restarts a failed host's machine on a healthy one and credits the interrupted session, and a missed spot notice credits 30 minutes. On a marketplace the hours lost to a host that drops offline, or to a Max-Q sold as a generic card, are usually yours. The scenarios below keep the disk all month, and where a storage price could not be read it is set to zero, which favours that provider. ### Three workloads, priced for a month **GPUBank** 730 h GPU $839.50 Storage $6.00 Other None Month $845.50 On demand, traffic included All-in **GPUBank** 730 h GPU $671.60 Storage $6.00 Other None Month $677.60 Reserved by the month, prepaid All-in p **packet.ai** 730 h GPU $481.80 Storage $0.00 Other Storage price not published Month $481.80 Shared card, no egress fee **Vast.ai** 730 h GPU $691.31 Storage $26.66 Other $7.81 traffic Month $725.78 Max-Q listing at $0.947/hr exact, 2 TB out at $0.0039/GB **Hyperstack** 730 h GPU $949.00 Storage $14.13 Other None Month $963.13 Reserved starting price **Runpod** 730 h GPU $1,817.70 Storage $20.00 Other None Month $1,837.70 Secure Cloud, volume disk GPUBank's month reservation is the cheapest dedicated card here: $671.60 for the GPU and $6.00 for 200 GB of disk make $677.60, against $725.78 for a Vast.ai Max-Q listing once its $26.66 of disk and $7.81 of traffic are added, $963.13 at Hyperstack's reserved starting price and $1,837.70 on Runpod's Secure Cloud. At on-demand list price GPUBank's $845.50 loses to the Vast.ai listing, and paid from $1,000 top-ups the same on-demand month costs $528.44 in cash. Only packet.ai's shared card undercuts the reservation, at $481.80 before a storage price it does not publish, and only if a card shared with other tenants gives the latency the API needs. The answer changes for a buyer who can only pay by card, for whom packet.ai or Vast.ai lead, and for anyone who accepts a 300 W Max-Q on a PCIe Gen 4 host, which runs a 4-bit 70B model at the same bandwidth but loses on prefill. 07 / Measured speed ## What it is fast at, measured The card's place follows from two numbers, 96 GB of memory and 1.6 to 1.8 TB/s of bandwidth. The memory holds a 70B model in 4-bit with tens of gigabytes left for KV cache, the same model in FP8 with a short cache, gpt-oss-120b in its native MXFP4, and Wan 2.2's 14B video experts without offloading. The bandwidth is roughly that of an RTX 5090 and about half of an H100 SXM, so single-user generation on a large model is modest: Hardware Corner measured Llama 3.3 70B in Q4_K with llama.cpp at 32.0 tokens per second at 4K of context, 28.2 at 16K and 16.6 at 128K. Batched serving is where the card pays: CloudRift's vLLM runs with 1,000-token prompts and answers at a concurrency of 400 gave one RTX PRO 6000 1,031 output tokens per second on Llama 3.3 70B AWQ, against 1,230 for two RTX 5090s and 467 for two RTX 4090s, and 8,425 on Qwen3-Coder-30B AWQ. Engine and precision move these figures a long way: NVIDIA's own TensorRT-LLM table, as reported by Eurokommerz, gives 1,724 output tokens per second per Server Edition on Llama 3.3 70B in FP4 with 1,000-token prompts and 296 with 8,192-token prompts, and StorageReview measured 163.1 tokens per second for one request on gpt-oss-120b in LM Studio on a Workstation card. On images, Naveen Kulandaivelu measured FLUX.1 dev at 1024 x 1024 at 3.10 iterations per second over 50 steps on a 600 W Workstation card, about 16.1 seconds per image. On video, HOSTKEY's tests of the Server Edition in ComfyUI put Wan 2.2 TI2V-5B at 160 seconds for a five-second 720p clip, and the 14B text-to-video model at about 40 minutes per clip in the card's default mode and 30 minutes after switching it to 600 W; HOSTKEY did not publish steps or precision. At GPUBank's $0.000319 per second that is $0.31 per million output tokens for batched 70B serving ($0.19 at the effective $0.72 hour), $11.33 per million for one user at 28 tokens per second, $0.05 per 5B clip, $0.58 to $0.77 per 14B clip and $5.15 per thousand FLUX images. The card stops being the right tool for multi-GPU training, which wants NVLink, for latency-critical 70B serving, where an H100 or H200 is faster per user, and for any model that fits in 32 GB. **Batched LLM serving on one card** Qwen3-Coder-30B-A3B AWQ, vLLM bench serve, 1,000 in and 1,000 out tokens, 1,200 requests, concurrency 400, FP8 KV cache RTX 4090 2,259 tok/s RTX 5090 4,570 tok/s RTX PRO 6000 8,425 tok/s CloudRift, 9 October 2025, one GPU per run on its own hosts. **Llama 3.3 70B in 4-bit: one card against two** Meta-Llama-3.3-70B-Instruct AWQ-INT4, vLLM, same settings as above, aggregate output throughput 2x RTX 4090 467 tok/s 1x RTX PRO 6000 1,031 tok/s 2x RTX 5090 1,230 tok/s CloudRift, 9 October 2025. The 70B model does not fit on one 32 GB or 24 GB card. **One user on a 70B model, by context length** Llama 3.3 70B Q4_K, llama.cpp, token generation 4K context 32 tok/s 16K context 28.2 tok/s 32K context 25.7 tok/s 64K context 21.7 tok/s 128K context 16.6 tok/s Hardware Corner, RTX PRO 6000 Blackwell; build, OS and power settings not published. CHEAPEST HOUR $0.33 Vast.ai interruptible bid on a Max-Q card GPUBANK ALL-IN $1.15 $0.72 effective after a $1,000 top-up BREAK-EVEN 523 h a month of use before a $16,000 card beats renting 08 / Per job, not per hour ## The RTX PRO 6000 against the other cards, per job On a model that fits in 32 GB, the RTX 5090 wins per job. CloudRift's Qwen3-Coder-30B AWQ run gave 4,570 tokens per second on one RTX 5090 against 8,425 on the RTX PRO 6000; at GPUBank's $0.48 and $1.15 an hour that is $0.029 per million tokens on the 5090 and $0.038 on the RTX PRO 6000. Trooper.ai's automated runs tell the same story on images, 8.3 seconds per SD3.5 Large image against 12 on the 5090, which is $0.0027 against $0.0016 per image at the same prices. The larger card earns its price only when the model would otherwise be offloaded, quantised harder or split across cards. Llama 3.3 70B in 4-bit, about 41 GB of weights, does not fit on one 5090, and CloudRift's two-5090 setup at $0.96 an hour on GPUBank produced 1,230 tokens per second against 1,031 for one RTX PRO 6000 at $1.15: close to a tie per token, with the single card keeping 50 GB free for cache and avoiding pipeline parallelism. The L40S at $0.55 has 48 GB and 864 GB/s, enough for a 4-bit 70B model with almost no room for context, and the only shared test we found is Runpod's single-user vLLM run, where the L40S decoded 48 tokens a second against 89 on the RTX PRO 6000. Against the 80 GB datacenter cards the measurements are few, and they split by workload. CloudRift ran GLM-4.5-Air at 4-bit under heavy batched load and measured 3,140 tokens a second on one RTX PRO 6000 against 2,987 on one H100, so at high concurrency the cheaper card holds its own. One user at a time is a different matter: Runpod's vLLM test at one request per second gave 89 tokens a second on the RTX PRO 6000, 96 on the A100 SXM, 109 on the H100 PCIe and 165 on the H100 SXM, which at GPUBank's prices is $3.59 per million tokens against $2.75, $3.44 and $3.11. Spread over several cards the lack of NVLink shows: in the same CloudRift study four H100s served Qwen3-Coder-480B about 31 percent faster than four RTX PRO 6000s, and eight H100s ran GLM-4.6 in FP8 nearly three times as fast as eight of them. For a team serving one large model to many users, or generating video, the RTX PRO 6000's 16 extra gigabytes and lower price usually decide it; for multi-GPU work or one-user latency on a 70B model, the H100 SXM at $1.85 is the better hour. RTX 5090 **32 GB** GDDR7 · 1.8 TB/s $0.48 /hr FLUX, SDXL, AI video and small language models L40S **48 GB** GDDR6 ECC · 864 GB/s $0.55 /hr Stable Diffusion, video, Whisper transcription and embeddings RTX PRO 6000 **96 GB** GDDR7 ECC · 1.8 TB/s $1.15 /hr Serving 70B models in 4-bit and AI video generation H100 PCIe **80 GB** HBM2e · 2.0 TB/s $1.35 /hr Single-GPU training and inference on a budget H100 SXM **80 GB** HBM3 · 3.35 TB/s $1.85 /hr AI training, fine-tuning and production LLM inference GPUBank on-demand prices, all-in. Memory bar scaled to 96 GB. 09 / Rent or buy ## Rent or buy one card Buying has become expensive. The Workstation Edition was listed at $8,565 at launch, NVIDIA's own US marketplace raised it from $13,250 to $16,000 in August 2026, against street prices in mid-2026 of $12,099.99 at Newegg for the Workstation Edition, $14,999 for the Server Edition and $13,999 at B&H for the Max-Q, and GPU Poet's October average of the three cheapest eBay listings each day was $16,963; Max-Q and Server Edition cards were trading higher still, at $22,256 and $23,288 on the same tracker. We take one Workstation card at $16,000 in a $4,000 workstation, drawing 0.8 kW at the wall under load, at $0.1819 per kWh, with no hosting fee and a resale value after 24 months of $6,400 for the card and $1,000 for the host. The host, power and resale figures are our assumptions, editable in the calculator. On those terms owning costs $12,600 of net capital over 24 months, $525 a month before use, plus $0.146 per loaded hour. Against GPUBank's $1.15 on-demand hour the two meet at $525 divided by $1.004, about 523 hours a month, a card busy 72 percent of the time for two years. Against the $0.72 effective hour from $1,000 top-ups the break-even would be 914 hours, more than a month holds, so renting is cheaper at any utilisation. At 730 hours owning costs about $631, below the $839.50 on-demand month and the $671.60 reserved month at list, but above the $525.60 bonus-funded on-demand month. At the old $8,565 price the capital falls to about $339 a month and the on-demand break-even to about 338 hours. Buying makes sense for someone who will keep the card busy most of the day for two years and cannot prepay, who needs the card on site, or who can still buy near launch price; everyone else pays less by renting. Monthly cost of one RTX PRO 6000 Workstation card (1 GPU) against hours of use a month, 0 to 730. Owning: price + $4,000 host − $7,400 resale over 24 months, plus 0.8 kW at the wall under load. Rent: per GPU-hour × 1. | Hours a month | Rent at $1.15 | Rent at $0.72 effective | Own | Cheaper | | --- | --- | --- | --- | --- | | 100 | $115.00 | $72.00 | $539.55 | Rent | | 200 | $230.00 | $144.00 | $554.10 | Rent | | 400 | $460.00 | $288.00 | $583.21 | Rent | | 523 | $601.45 | $376.56 | $601.11 | Even | | 730 | $839.50 | $525.60 | $631.23 | Own, unless rent uses the bonus | NVIDIA US marketplace price since August 2026, per Igor's Lab (14 Aug 2026); GPU Poet's October 2026 eBay average was $16,963 10 / What to choose ## The cheapest RTX PRO 6000 depends on the job. Serving a 70B model or gpt-oss-120b to many users RTX PRO 6000, reserved by the month One card holds the 4-bit weights with 50 GB left for cache, and batched vLLM reached 1,031 tokens per second on Llama 3.3 70B; at $0.92 reserved that is the cheapest dedicated month we could confirm. Generating video with Wan 2.2 RTX PRO 6000, interruptible Clips are independent, so a reclaimed card costs one clip; 2,000 five-second 5B clips came to $70.33 a month on GPUBank's $0.69 tier, and the 14B experts fit without offloading. Images, LoRA and models under 32 GB RTX 5090 When the model fits in 32 GB the RTX 5090 at $0.48 does the job for less per output: $0.029 per million tokens against $0.038 on the 30B test. Training across several GPUs H100 SXM The RTX PRO 6000 has no NVLink and about half the H100 SXM's bandwidth; for multi-GPU training and one-user latency on large models the $1.85 H100 SXM is the better hour. 11 / Questions ## What people ask before renting one. The cheapest RTX PRO 6000 hour on 16 October 2026 was an interruptible Vast.ai bid at $0.33 on a single 300 W Max-Q card, which can be taken back whenever a higher bid arrives and carries its own disk charge. The cheapest non-interruptible marketplace listing was a Max-Q on Vast.ai at $0.95, with the cheapest full-power Workstation card at $1.00. packet.ai sells a shared card at $0.66 and HyperAI lists $0.80 without stating the machine. The cheapest dedicated card with a full machine on a provider's own page was Beam at $1.09, followed by GPUBank at $1.15 with traffic and IPv4 included, or $0.72 effective after the bonus on a $1,000 top-up. Which is cheapest for you depends on whether you accept interruption, a shared card or a 300 W edition, and whether you prepay. Sources ## Sources 1. 01 [Vast.ai public offer search, RTX PRO 6000 S, WS and Max-Q](https://console.vast.ai/api/v0/bundles/) Live rentable offers per edition, prices, minimum bids, disk and bandwidth rates (read 02:00 EEST) · read 16 Oct 2026 2. 02 [Beam pricing](https://www.beam.cloud/pricing) RTX PRO 6000 machine from $1.09/hr with 16 vCPU, 128 GB RAM, 1 TB NVMe; plans · read 16 Oct 2026 3. 03 [GPUBank pricing](https://gpubank.io/pricing) GPUBank RTX PRO 6000 rates in three modes, inclusions, disk price, billing rules, top-up bonuses, other cards · read 16 Oct 2026 4. 04 [GetDeploying, Nvidia RTX PRO 6000 rental prices](https://getdeploying.com/gpus/nvidia-rtx-pro-6000) Aggregator rows crawled 15 Oct 2026, median $2.21 across 165 on-demand configs · read 16 Oct 2026 5. 05 [Runpod pricing](https://www.runpod.io/pricing) RTX Pro 6000 $2.49 Secure Cloud, $1.69 Community Cloud, $3.49 serverless; disk rates · read 16 Oct 2026 6. 06 [Verda pricing](https://verda.com/pricing) RTX PRO 6000 $2.30 on demand, $1.15 spot, $2.53 serverless · read 16 Oct 2026 7. 07 [Nebius AI Cloud, compute pricing](https://docs.nebius.com/compute/resources/pricing) $1.80 on demand, preemptible from $0.79 under dynamic spot pricing since 8 Oct 2026 · read 16 Oct 2026 8. 08 [Vast.ai, Rent RTX PRO 6000 WS GPUs](https://vast.ai/pricing/gpu/RTX-PRO-6000-WS) Pricing page showing no current offers; launch price 8,565 USD in the specs · read 16 Oct 2026 9. 09 [NVIDIA RTX PRO 6000 Blackwell Workstation Edition](https://www.nvidia.com/en-us/products/workstations/professional-desktop-gpus/rtx-pro-6000/) 96 GB GDDR7 ECC, 1,792 GB/s, 600 W, double flow-through cooler, PCIe Gen 5 · read 16 Oct 2026 10. 10 [PNY, NVIDIA RTX PRO 6000 Blackwell Workstation Edition](https://www.pny.com/nvidia-rtx-pro-6000-blackwell-ws) CUDA, Tensor and RT core counts, 512-bit bus, MIG slices · read 16 Oct 2026 11. 11 [NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition](https://www.nvidia.com/en-us/products/workstations/professional-desktop-gpus/rtx-pro-6000-max-q/) 96 GB GDDR7 ECC, 1,792 GB/s, 300 W, dual slot · read 16 Oct 2026 12. 12 [NVIDIA RTX PRO 6000 Blackwell Server Edition](https://www.nvidia.com/en-us/data-center/rtx-pro-6000-blackwell-server-edition/) 96 GB GDDR7 ECC, 1,597 GB/s, up to 600 W configurable, 4 PFLOPS FP4 · read 16 Oct 2026 13. 13 [ServeTheHome, NVIDIA RTX PRO 6000 Blackwell family for workstations and servers](https://www.servethehome.com/nvidia-rtx-pro-6000-blackwell-family-for-workstations-and-servers/) Coolers of the three editions, Server Edition configurable from 400 to 600 W · read 16 Oct 2026 14. 14 [NVIDIA GeForce software licence](https://www.nvidia.com/en-us/drivers/geforce-license/) No datacenter deployment of GeForce cards · read 15 Oct 2026 15. 15 [AWS, Amazon EC2 G7e instances](https://aws.amazon.com/ec2/instance-types/g7e/) G7e sizes with RTX PRO 6000 Blackwell Server Edition · read 16 Oct 2026 16. 16 [Economize, GCP g4-standard-48 pricing](https://www.economize.cloud/resources/gcp/pricing/compute-engine/g4-standard-48/) us-central1: $4.50 on demand, $3.105 1-year, $1.979 3-year, $1.715 spot · read 16 Oct 2026 17. 17 [Oracle Cloud Infrastructure, compute shapes](https://docs.oracle.com/en-us/iaas/Content/Compute/References/computeshapes.htm) BM.GPU.RTXPRO.8: 8 GPUs, 144 OCPU, 3,072 GB RAM, 8 x 7.68 TB NVMe · read 16 Oct 2026 18. 18 [CoreWeave pricing](https://www.coreweave.com/pricing) 8-GPU RTX PRO 6000 Server Edition at $20.00/h on demand, $9.56 and $11.09 spot · read 16 Oct 2026 19. 19 [Hyperstack GPU pricing](https://www.hyperstack.cloud/gpu-pricing) RTX PRO 6000 SE $1.85 on demand, $1.48 spot, from $1.30 reserved; volume storage rate · read 16 Oct 2026 20. 20 [packet.ai pricing](https://packet.ai/pricing) RTX 6000 Pro at $0.66/hr on the multi-tenant Dynamic tier, hourly billing, no egress fees · read 16 Oct 2026 21. 21 [HyperAI pricing](https://hyper.ai/en/pricing) RTX PRO 6000 at $0.80/h · read 16 Oct 2026 22. 22 [Clore.ai documentation, GPU pricing and availability](https://docs.clore.ai/guides/getting-started/pricing) Snapshot of 24 Aug 2026: $0.92 to $1.38, median $1.21, 4 free servers · read 16 Oct 2026 23. 23 [Lium, RTX PRO 6000 Blackwell Server Edition](https://lium.io/gpu/rtx-pro-6000-blackwell-server-edition) Live asks $1.60 to $1.99, 2 pods, billed per second · read 16 Oct 2026 24. 24 [Verda, NVIDIA RTX PRO 6000](https://verda.com/rtx-pro-6000) Server Edition, 30 CPU threads and 90 GB RAM per GPU · read 16 Oct 2026 25. 25 [Massed Compute pricing](https://vm.massedcompute.com/pricing) RTX PRO 6000 Blackwell at $2.19 with 16 vCPU, 144 GB RAM, 725 GB · read 16 Oct 2026 26. 26 [CloudRift pricing](https://www.cloudrift.ai/pricing) RTX PRO 6000 at $1.16 for one month and $1.10 for three months · read 16 Oct 2026 27. 27 [Vantage, g7e.2xlarge pricing](https://instances.vantage.sh/aws/ec2/g7e.2xlarge) $3.363/h on demand, $2.557/h spot · read 16 Oct 2026 28. 28 [Vantage, g7e.48xlarge pricing](https://instances.vantage.sh/aws/ec2/g7e.48xlarge) $33.144/h on demand for 8 GPUs · read 16 Oct 2026 29. 29 [GPUBank home page](https://gpubank.io/) Regions listing the RTX PRO 6000, median launch time, templates · read 16 Oct 2026 30. 30 [Vast.ai documentation, Billing](https://docs.vast.ai/documentation/reference/billing) Storage billed while an instance is stopped, bandwidth billed per byte · read 15 Oct 2026 31. 31 [OpenAI, gpt-oss-120b model card](https://huggingface.co/openai/gpt-oss-120b) 117B parameters, MXFP4 MoE weights, fits one 80 GB GPU · read 16 Oct 2026 32. 32 [Wan-AI, Wan2.2-T2V-A14B model card](https://huggingface.co/Wan-AI/Wan2.2-T2V-A14B) 27B total parameters, 14B active, 80 GB single-GPU command · read 16 Oct 2026 33. 33 [Hardware Corner, RTX PRO 6000 Blackwell LLM benchmarks](https://www.hardware-corner.net/gpu-llm-benchmarks/rtx-pro-6000-blackwell/) llama.cpp generation and prompt speeds by context length · read 16 Oct 2026 34. 34 [CloudRift, RTX 4090 vs 5090 vs PRO 6000 LLM inference benchmark](https://www.cloudrift.ai/blog/benchmarking-rtx-gpus-for-llm-inference) vLLM batched throughput on Qwen3-Coder-30B AWQ and Llama 3.3 70B AWQ (9 Oct 2025) · read 16 Oct 2026 35. 35 [Eurokommerz, RTX PRO 6000 Blackwell benchmarks and MLPerf results](https://eurokommerz.at/blog/nvidia-rtx-pro-6000-real-benchmarks/) NVIDIA TensorRT-LLM table: 1,724 and 296 tok/s on Llama 3.3 70B FP4 (24 Sep 2026) · read 16 Oct 2026 36. 36 [StorageReview, NVIDIA RTX PRO 6000 Workstation GPU review](https://www.storagereview.com/review/nvidia-rtx-pro-6000-workstation-gpu-review-blackwell-architecture-and-96-gb-for-pro-workflows) 163.1 tok/s on gpt-oss-120b, one request, LM Studio · read 16 Oct 2026 37. 37 [Naveen Kulandaivelu, Benchmarking Nvidia's RTX PRO 6000](https://www.naveen.ing/writings/benchmarking-rtx-pro-6000/) FLUX.1 dev 3.10 it/s at 1024 x 1024, power configurable down to 150 W (7 Aug 2025) · read 16 Oct 2026 38. 38 [HOSTKEY, RTX 6000 Blackwell Server Edition tests and benchmarks](https://hostkey.com/blog/109-nvidia-rtx-6000-blackwell-server-edition-tests-benchmarks-comparison-with-workstation-and-rtx-5090-cooling-features/) Wan 2.2 14B and TI2V-5B times, 25 percent gain in 600 W mode (28 Aug 2025) · read 16 Oct 2026 39. 39 [Trooper.ai, RTX Pro 6000 Blackwell vs RTX 5090 benchmark](https://www.trooper.ai/benchmarks-compare-RTX-Pro-6000-Blackwell-with-RTX-5090) SD3.5 Large 8.3 s against 12 s per image · read 16 Oct 2026 40. 40 [Runpod, Best GPU for LLM inference: what the benchmarks show in 2026](https://www.runpod.io/articles/guides/best-gpu-for-llm-inference) vLLM single-user decode: RTX PRO 6000 89, L40S 48, A100 SXM 96, H100 PCIe 109, H100 SXM 165 tok/s (measured 15 Sep 2026) · read 16 Oct 2026 41. 41 [CloudRift, RTX PRO 6000 vs H100, H200 and L40S: LLM inference](https://www.cloudrift.ai/blog/benchmarking-rtx6000-vs-datacenter-gpus) GLM-4.5-Air AWQ 3,140 against 2,987 tok/s; multi-GPU gaps on Qwen3-Coder-480B and GLM-4.6 (27 Nov 2025) · read 16 Oct 2026 42. 42 [Tom's Hardware, Nvidia raises RTX Pro 6000 Blackwell pricing to $13,250](https://www.tomshardware.com/pc-components/gpus/nvidia-raises-rtx-pro-6000-blackwell-gpu-pricing-to-usd13-250-55-percent-increase-over-msrp-in-a-years-time) Mid-2026 street prices at Newegg and B&H for the three editions · read 16 Oct 2026 43. 43 [Igor's Lab, RTX PRO 6000 Blackwell now costs 16,000 US dollars](https://www.igorslab.de/en/nvidia-rtx-pro-6000-blackwell-16000-dollar-96-gb-workstation-gpu-price-increase/) Workstation Edition at $16,000 on NVIDIA's US marketplace, up from $13,250 (14 Aug 2026) · read 16 Oct 2026 44. 44 [GPU Poet, RTX PRO 6000 Workstation Edition price, October 2026](https://gpupoet.com/gpu/learn/price/october-2026/nvidia-rtx-pro-6000-blackwell-workstation) Lowest average of the three cheapest eBay listings: $16,963 ($16,325 to $17,860) · read 16 Oct 2026 45. 45 [GPU Poet, RTX PRO 6000 Max-Q price, October 2026](https://gpupoet.com/gpu/learn/price/october-2026/nvidia-rtx-pro-6000-blackwell-max-q) Lowest average $22,256, estimated MSRP $10,999 · read 16 Oct 2026 46. 46 [GPU Poet, RTX PRO 6000 Server Edition price, October 2026](https://gpupoet.com/gpu/learn/price/october-2026/nvidia-rtx-pro-6000-blackwell-server) Lowest average $23,288, cheapest listing $22,999, no official MSRP · read 16 Oct 2026 47. 47 [Puget Systems, RTX PRO 6000 Blackwell Max-Q vs Workstation for content creation](https://www.pugetsystems.com/labs/articles/nvidia-rtx-pro-6000-blackwell-max-q-vs-workstation-for-content-creation/) Max-Q 5 to 14 percent slower than the 600 W card in most tests (July 2025) · read 16 Oct 2026 48. 48 [Beam, NVIDIA RTX PRO 6000 Pricing](https://www.beam.cloud/blog/nvidia-rtx-pro-6000-pricing) $1.09/hr with vCPU, RAM and NVMe included (29 Sep 2026) · read 16 Oct 2026 What we could not confirm We could not read Hostinger's, GPUhub's, Nova Cloud's, UpCloud's, Leaseweb's, Database Mart's, Utho's, Azure's or Oracle's RTX PRO 6000 prices on their own pages, so those rows come from GetDeploying's crawl of 15 October 2026, and the AWS and Google prices come from Vantage and Economize because the providers' price pages did not render. HyperAI's pricing page shows $0.80 but not the machine behind it, and its $0.48 one-month reservation appears only on GetDeploying. Clore.ai's latest own figure is its snapshot of 24 August 2026. Vast.ai's prices come from its public offer search at about 02:00 Tallinn time on 16 October, because its pricing pages showed no offers; bids and listings change by the hour. Beam does not say whether $1.09 is an exact or starting rate, nor which edition it rents. GPUBank's pricing page does not name the edition or power limit behind its RTX PRO 6000 rentals. Aurora's widely quoted 33 percent Max-Q gap on Wan 2.2 could not be read and was left out in favour of Puget Systems' measurements. We found no measured HunyuanVideo or LTX-Video times on this card, no single-user benchmark beyond Runpod's that puts an RTX PRO 6000 and an H100 or L40S on the same test, NVIDIA's TensorRT-LLM table only as Eurokommerz reports it, and no dated price for a complete 8-GPU server. The host price, power draw and resale values in the rent-or-buy arithmetic are assumptions. RTX PRO 6000 on GPUBank ## 96 GB on one card, with the whole machine included. - 16 vCPU, 188 GB RAM, 2 TB NVMe, IPv4 and traffic in $1.15/hr - Billed by the second after the health check - Interruptible $0.69, reserved $0.92 [Launch an RTX PRO 6000 · $1.15/hr](https://console.gpubank.io/launch?gpu=rtx-pro-6000) [Compare 63 GPU clouds](https://gpubank.io/compare) More GPU price guides ## Where to rent the other cards at the cheapest price. [B200 SXM 180 GB HBM3e From $2.49 spot · $4.15 all-in](https://gpubank.io/rent-b200) [H200 SXM 141 GB HBM3e From $1.49 spot · $2.49 all-in](https://gpubank.io/rent-h200) [H100 SXM 80 GB HBM3 From $1.11 spot · $1.85 all-in](https://gpubank.io/rent-h100) [H100 PCIe 80 GB HBM2e From $0.81 spot · $1.35 all-in](https://gpubank.io/rent-h100-pcie) [MI300X 192 GB HBM3 From $0.99 spot · $1.65 all-in](https://gpubank.io/rent-mi300x) [A100 SXM 80 GB HBM2e From $0.57 spot · $0.95 all-in](https://gpubank.io/rent-a100) [L40S 48 GB GDDR6 ECC From $0.33 spot · $0.55 all-in](https://gpubank.io/rent-l40s) [RTX 5090 32 GB GDDR7 From $0.29 spot · $0.48 all-in](https://gpubank.io/rent-rtx-5090) [RTX 4090 24 GB GDDR6X From $0.14 spot · $0.23 all-in](https://gpubank.io/rent-rtx-4090)