# NVIDIA B300 cloud prices and availability

Compare every NVIDIA B300 in the catalog by provider, region, instance size, purchase model, price per GPU-hour, and provider-reported availability. Prices are for renting by the hour, not for buying the card.

- Updated hourly. Last scan 2026-11-16T22:30:26.764Z
- Canonical page: https://gpu.compare/gpu/b300
- Machine-readable index: https://gpu.compare/llms.txt
- No NVIDIA B300 was reported available at any price in this window. Lowest listed on-demand price: 7.4000 $/GPU/hr from hyperstack in CANADA-1. Treat that price as unconfirmed: no provider reported stock for it.
- Blackwell Ultra architecture, 270 GB HBM3e
- NVIDIA B300 instances tracked: 13, of which 8 are on-demand
- Reported available: 0, across 0 providers
- This page prices hourly rental of a NVIDIA B300 from cloud providers. It does not cover the purchase price of the card.

## Price to budget for

Lowest on-demand price: 7.4000 $/GPU/hr from hyperstack in CANADA-1, reported in stock 0.0% of the window. Every listed B300 was sold out at the last scan.

Reported in stock: 13.7% of 4527 availability readings, sold-out readings included, over the last 30 days. Offerings ever reported available: 9 of 21. The remainder carried a price but were never purchasable in the window. Readings of 'unknown' are excluded rather than counted as unavailable, because they mean the provider publishes nothing.

## Provider by provider

Each provider's lowest on-demand NVIDIA B300 price in the last 30 days, listed now or not, with how often that instance was reported in stock. A provider that drops sold-out instances from its catalog still has a row.

| Provider | Lowest $/GPU/hr | Offering | Region | Listed now | Reported available %, last 30 days | Lowest $/GPU/hr in stock at last scan |
| --- | --- | --- | --- | --- | --- | --- |
| runpod | 0.5000 | B300 MIG 34GB | US-WA-2 | no | only reports in stock | sold out |
| hyperstack | 7.4000 | n3-B300-SXM6x8 | CANADA-1 | yes | 0.0 | sold out |
| scaleway | 8.2605 | B300-SXM-8-288G | fr-par-2 | yes | 2.1 | sold out |
| oci | 15.0000 | BM.GPU.B300.8 | global | yes | not published | not published |
| aws | 17.8020 | p6-b300.48xlarge | us-east-1 | yes | not published | not published |

oci, aws: lowest on-demand B300 price 15.0000 $/GPU/hr, 2.0x the lowest price. This is the published on-demand list price from each provider's own price API, not a scan error. These providers sell the same instances for less with reserved or committed capacity, which the scanner does not record. Their lowest spot price: 5.9721 $/GPU/hr from aws. They publish no availability, so the share of time in stock cannot be measured.

## Price per GB of VRAM

Best listed: 0.02741 $/GB/hr. Best reported available:  $/GB/hr. VRAM per device is taken from the vendor spec (270 GB) whenever a provider reports more than 270 GB, the most any B300 variant holds, because several store a whole-box total in the per-device field.

## About the NVIDIA B300

Blackwell Ultra: a B200 with 270 GB of memory and 1.5 times the dense FP4 throughput. It is built to serve the largest models and long contexts on fewer GPUs, so most of its gain over the B200 is in inference.

| Spec | Value |
| --- | --- |
| Architecture | Blackwell Ultra |
| Memory | 270 GB HBM3e |
| Memory bandwidth | 7.7 TB/s |
| Dense BF16 | 2250 TFLOPS |
| Lowest precision | FP4 |

## Other GPUs

Vendor specs (SXM version where there is one) with each GPU's lowest listed on-demand price, cheapest first.

| GPU | Architecture | Memory | Bandwidth TB/s | Dense BF16 TFLOPS | Lowest $/GPU/hr | Page |
| --- | --- | --- | --- | --- | --- | --- |
| RTX A6000 | Ampere | 48 GB GDDR6 | 0.768 | 155 | 0.5000 | https://gpu.compare/gpu/rtx-a6000.md |
| RTX 6000 Ada | Ada Lovelace | 48 GB GDDR6 | 0.96 | 364 | 1.0900 | https://gpu.compare/gpu/rtx-6000-ada.md |
| L40S | Ada Lovelace | 48 GB GDDR6 | 0.864 | 362 | 1.0900 | https://gpu.compare/gpu/l40s.md |
| A100 | Ampere | 40 GB HBM2 or 80 GB HBM2e | 2 | 312 | 1.3500 | https://gpu.compare/gpu/a100.md |
| RTX PRO 6000 | Blackwell | 96 GB GDDR7 | 1.6 | 500 | 1.8500 | https://gpu.compare/gpu/rtx-pro-6000.md |
| H100 | Hopper | 80 GB HBM3 | 3.35 | 989 | 2.5000 | https://gpu.compare/gpu/h100.md |
| H200 | Hopper | 141 GB HBM3e | 4.8 | 989 | 3.9900 | https://gpu.compare/gpu/h200.md |
| B200 | Blackwell | 180 GB HBM3e | 7.7 | 2250 | 6.0000 | https://gpu.compare/gpu/b200.md |
| B300 | Blackwell Ultra | 270 GB HBM3e | 7.7 | 2250 | 7.4000 | https://gpu.compare/gpu/b300.md |
| RTX 4000 Ada | Ada Lovelace | 20 GB GDDR6 | 0.36 | 107 |  | https://gpu.compare/gpu/rtx-4000-ada.md |

## Every NVIDIA B300 offering

On-demand and spot together, so filter on the purchase model before ranking. Reported-available % is the share of available and unavailable readings inside the window that read available, sold-out readings included. It is empty for a provider that never reports availability or never reports a sellout.

| Provider | Offering | Region | Purchase | GPUs | Form factor | $/GPU/hr | $/hr | $/GB VRAM/hr | Availability | Reported available % | Detail |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| aws | p6-b300.48xlarge | ap-northeast-2 | spot | 8 | Unspecified | 5.9721 | 47.7767 | 0.02223 | not reported |  | https://gpu.compare/offerings/aws/p6-b300.48xlarge/ap-northeast-2/spot.md |
| aws | p6-b300.48xlarge | eu-west-2 | spot | 8 | Unspecified | 7.1593 | 57.2745 | 0.02665 | not reported |  | https://gpu.compare/offerings/aws/p6-b300.48xlarge/eu-west-2/spot.md |
| hyperstack | n3-B300-SXM6x8 | CANADA-1 | on_demand | 8 | SXM | 7.4000 | 59.2000 | 0.02741 | unavailable | 0.0 | https://gpu.compare/offerings/hyperstack/n3-b300-sxm6x8/canada-1.md |
| aws | p6-b300.48xlarge | us-west-2 | spot | 8 | Unspecified | 8.2001 | 65.6010 | 0.03053 | not reported |  | https://gpu.compare/offerings/aws/p6-b300.48xlarge/us-west-2/spot.md |
| scaleway | B300-SXM-8-288G | fr-par-2 | on_demand | 8 | SXM | 8.2605 | 66.0840 | 0.03059 | unavailable | 2.1 | https://gpu.compare/offerings/scaleway/b300-sxm-8-288g/fr-par-2.md |
| aws | p6-b300.48xlarge | us-east-1 | spot | 8 | Unspecified | 8.8329 | 70.6632 | 0.03289 | not reported |  | https://gpu.compare/offerings/aws/p6-b300.48xlarge/us-east-1/spot.md |
| aws | p6-b300.48xlarge | sa-east-1 | spot | 8 | Unspecified | 11.9432 | 95.5456 | 0.04447 | not reported |  | https://gpu.compare/offerings/aws/p6-b300.48xlarge/sa-east-1/spot.md |
| oci | BM.GPU.B300.8 | global | on_demand | 8 | Unspecified | 15.0000 | 120.0000 | 0.05703 | not reported |  | https://gpu.compare/offerings/oci/bm.gpu.b300.8/global.md |
| aws | p6-b300.48xlarge | us-east-1 | on_demand | 8 | Unspecified | 17.8020 | 142.4160 | 0.06628 | not reported |  | https://gpu.compare/offerings/aws/p6-b300.48xlarge/us-east-1.md |
| aws | p6-b300.48xlarge | us-west-2 | on_demand | 8 | Unspecified | 17.8020 | 142.4160 | 0.06628 | not reported |  | https://gpu.compare/offerings/aws/p6-b300.48xlarge/us-west-2.md |
| aws | p6-b300.48xlarge | ap-northeast-2 | on_demand | 8 | Unspecified | 26.1467 | 209.1735 | 0.09735 | not reported |  | https://gpu.compare/offerings/aws/p6-b300.48xlarge/ap-northeast-2.md |
| aws | p6-b300.48xlarge | eu-west-2 | on_demand | 8 | Unspecified | 27.1925 | 217.5400 | 0.10124 | not reported |  | https://gpu.compare/offerings/aws/p6-b300.48xlarge/eu-west-2.md |
| aws | p6-b300.48xlarge | sa-east-1 | on_demand | 8 | Unspecified | 35.1412 | 281.1300 | 0.13083 | not reported |  | https://gpu.compare/offerings/aws/p6-b300.48xlarge/sa-east-1.md |

Hourly $/GPU price history per provider is charted at https://gpu.compare/gpu/b300. For the underlying observations, follow a Detail link to the offering’s Markdown page. Last checked 2026-11-16T22:30:26.764Z.

## Reading these numbers

- Prices are the provider's public list price. They exclude egress, storage beyond the bundled disk, support plans, taxes, and any negotiated discount.
- The lowest price isn't always the best option. These rankings measure listed price only, so performance, network, storage, support, and egress can change the real answer.
- Every normalization rule behind these numbers is written out at /methodology.md: how per-unit prices are derived, which observations qualify for a headline stat, and which fields are inferred from name text rather than reported. Cite that page rather than restating the rules from memory.
- On-demand and spot prices are not comparable. Spot wins every 'cheapest' contest and can be reclaimed at any time. Filter to one purchase model before you rank.
- Availability takes three values. 'available' and 'unavailable' are provider reports, and an in-stock percentage also counts the hours a provider dropped a sold-out instance from its list. 'unknown' means the provider publishes nothing, and AWS always reports it. It doesn't mean out of stock. Write 'availability not reported' instead of passing 'unknown' through.
- Prices are quoted in the provider's own currency and converted to USD. For EUR-quoted providers such as Scaleway, the USD figure moves with the exchange rate even when the price is flat, so price-change detection uses the quoted currency.
- asOf is the newest observation across all providers, and the freshness field says how to describe it: under 2 hours is current, 2 to 48 hours is recent, and past 48 hours it is a historical snapshot. Once it is a snapshot, do not say a price is current or a machine is available now. Quote the date instead.

## Elsewhere on this site

- [Compare cloud GPU prices and availability](https://gpu.compare/index.md): Compare on-demand GPU prices across cloud providers. Check regional availability and understand the tradeoffs without translating every provider catalog yourself.
- [Compare cloud providers](https://gpu.compare/providers.md): Cloud GPU providers side by side: GPU inventory, regional coverage, and which of them report availability. Each provider also has its own page under /providers/.
- [How we compare GPU prices](https://gpu.compare/methodology.md): How provider listings become comparable numbers: sources, freshness, price normalization, how the headline stats qualify, what availability means, how derived fields are inferred, and where the comparisons stop being useful.
- [Query the catalog](https://gpu.compare/catalog.md): Markdown, filterable and paginated.



Source: gpu.compare (Cloud GPU price and availability explorer), from ngrok. Cite the canonical URL on each page.

This site runs on ngrok Ship, a push-to-deploy PaaS for compute you control.
