Log in Rent B300
B300 Blackwell Ultra · 720-day term · 288 GB a card

Rent a B300:
720 days minimum.
There is a 30-day tier.

This is the longest commitment on Clore.ai and the largest card on it. Blackwell Ultra carries 288 GB of HBM3e at 8,000 GB/s and draws 1,400 W. Clore contracts it at $5.02 to $5.16 per GPU-hour, minimum 8 GPUs, minimum 720 days, in the USA, the EU and Japan as of 21 Sep 2026. Every longer tier is priced the same, so the term buys certainty rather than a discount. Zero B300 servers were listed on the per-minute marketplace that day.

720-day minimum term 8 GPUs minimum USA, EU, Japan 1,400 W a card
288GB
HBM3e per GPU
720days
Shortest term offered
1,400W
Board power per GPU
$5.02/GPU-hr
EU rate, 21 Sep 2026

A line item,
not a rental.

Every other card on Clore.ai can be had for a month. This one cannot, and that changes which question you should be asking first.

The term decides before the hardware does

Eight cards at the EU rate for the minimum term is roughly $694,000 of committed spend, and at the USA and Japan rate roughly $713,000. Clore prices 720, 1,080, 1,440 and 1,800 days identically, so extending the commitment buys certainty of supply and nothing off the rate. Work out whether you want this silicon for two years before you read another line of this page.

Rate at 720 through 1,800 days identical

The largest per-GPU capacity Clore contracts

288 GB moves several models out of the tensor-parallel category entirely. Llama 3.1 405B at INT4 is roughly 200 GB of weights and lands on one card. DeepSeek-V3's 671B mixture of experts is about 336 GB at FP4, so two. What that buys is not speed, it is the absence of a collective at every layer.

Llama 3.1 405B at INT4 one card

Blackwell Ultra gains in exactly one place

NVIDIA's HGX page gives the eight-GPU B300 system 144 PFLOPS of FP4 with sparsity and 108 PFLOPS dense, where HGX B200 is 144 and 72. FP8, FP6, FP16, BF16 and TF32 are identical on both platforms. The generational step is dense four-bit throughput, half again as much, and nothing else.

Dense FP4 over HGX B200 1.5×

What each contract
asks of you.

Rates and minimums as the bare-metal configurator returned them on 21 Sep 2026. The row worth staring at is the last one.

B300 B200 H200 H100
VRAM per GPU 288 GB HBM3e 180 GB HBM3e 141 GB HBM3e 80 GB HBM3
Board power 1,400 W 1,000 W 700 W 700 W
Minimum GPUs 8 8 8 8
Minimum term 720 days 30 days 30 days 30 days
Rate at that minimum $5.02-$5.16 $5.59 $3.05-$4.18 $2.38-$2.62
Rate at the longest term $5.02-$5.16 $3.93 $2.09 $1.81
What a longer term saves nothing about 30% about 33% about 31%

// Customer-facing rates per GPU-hour from Clore's bare-metal route, 21 Sep 2026. Savings are the drop from each card's shortest tier to its longest, measured on that card's highest-priced entry.

Three regions,
fourteen cents between them.

B300 has no term ladder, so region is the only variable that moves the price. Over eight cards and the minimum term the gap works out at roughly $19,000.

USA and Japan

$5.16 / GPU-hr
Flat from 720 to 1,800 days · 21 Sep 2026
  • Two separate entries, USA and Japan, at the same rate
  • About $713,000 for 8 cards over the 720-day minimum
  • Pod sizes from 8 to 1,000 GPUs
  • No shorter tier exists in either region
Price USA or Japan
LOWER RATE

EU

$5.02 / GPU-hr
Flat from 720 to 1,800 days · 21 Sep 2026
  • The cheapest B300 rate Clore quotes
  • About $694,000 for 8 cards over the 720-day minimum
  • Listed as a region rather than a country
  • Settled in BTC, CLORE, USDT or USDC
Price the EU
Invoiced against
Bitcoin on-chain
CLORE native token
USDT / USDC ERC-20 · BEP-20

Four questions, in this order.

A 720-day contract is a procurement exercise, not a checkout. The configurator above collects the inputs; the first decision is not one it can make for you.

01 / HORIZON

Do you want it for two years?

There is no 30-day tier, no 90 and no 360. If your horizon is shorter than 720 days, a B200 or H200 contract is the honest answer and both are on this site.

02 / SIZE

How many cards

Eight is the minimum and a thousand the maximum. Eight cards is one Blackwell Ultra chassis on an NVLink 5 fabric.

03 / REGION

EU, or the USA and Japan

$5.02 against $5.16 per GPU-hour. Across eight cards over the minimum term that difference is roughly $19,000, which may or may not outweigh where you need the data to sit.

8 × B300 · 720 days · EU
04 / TERMS

Send it and negotiate

Clore returns a contract quote. Invoicing runs against your BTC, CLORE, USDT or USDC balance, the same as every other order on the platform.

Questions a two-year term raises.

The B300 minimum term is 720 days. Who is that contract actually for?

Someone with a workload they already know the shape of. The commitment is roughly $694,000 in the EU or $713,000 in the USA and Japan for the minimum eight cards, before power, staff or anything else, and it does not get cheaper if you extend it. That suits an inference service with committed customers, a lab with a funded two-year research programme, or a company replacing a cloud bill it can already measure. It does not suit exploratory work, a single training run, or anyone who cannot say what they will be running in eighteen months.

288 GB per GPU: which model classes need that rather than 141 GB?

The ones where splitting the model is the cost you are trying to avoid. Llama 3.1 405B at INT4 is roughly 200 GB of weights: one B300 card, against two H200s. DeepSeek-V3's 671B mixture of experts is about 336 GB at FP4: two cards against three. Trillion-parameter-class mixtures of experts are the class this capacity exists for. If your largest model fits in 141 GB with cache to spare, the extra capacity is idle and you are paying for it every hour of both years.

B300 versus B200: the FP16 figure is the same. Where does Blackwell Ultra actually gain?

In dense FP4, and only there. NVIDIA's HGX platform page gives the eight-GPU B300 system 144 PFLOPS of FP4 with sparsity and 108 PFLOPS dense, against HGX B200's 144 and 72. FP8 and FP6 are 72 PFLOPS on both, FP16 and BF16 are 36 PFLOPS on both, TF32 is 18 PFLOPS on both. Per GPU, memory goes from 180 GB to 288 GB and board power from 1,000 W to 1,400 W. So the upgrade is capacity plus half again as much dense four-bit throughput, bought with 40 percent more power.

What does 288 GB × 8 GPUs let me hold resident that an H200 pod cannot?

NVIDIA lists 2.1 TB of total memory for the eight-GPU HGX B300 system against 1.1 TB for an eight-card H200 node at 141 GB each. In practice that is the difference between holding one frontier model plus a deep KV cache and holding two frontier models, or one model and the retrieval and reranking stack in front of it. Note that NVIDIA's 2.1 TB is below eight times the 288 GB part figure, so confirm the per-GPU capacity of the SKU you are contracting rather than multiplying.

Is there any route to B300 shorter than 720 days on CLORE?

No. The bare-metal configurator offers 720, 1,080, 1,440 and 1,800 days for B300 and nothing below, and the per-minute marketplace held zero B300 servers on 21 September 2026. If you need Blackwell for less than two years, the B200 contract starts at 30 days at $5.59 per GPU-hour, and if you need capacity rather than FP4 throughput, the H200 contract also starts at 30 days and is the cheapest of these per gigabyte held.

Why does NVIDIA's HGX B300 page show lower INT8 and FP64 throughput than the B200?

Because Blackwell Ultra rebalanced the die toward low-precision AI work. NVIDIA's HGX platform page lists INT8 tensor throughput at 3 POPS for the eight-GPU B300 system against 72 POPS for B200, and FP64 at 10 TFLOPS against 296. FP16, BF16, TF32 and FP8/FP6 are identical between them. So if your workload is INT8-quantised inference from before the FP4 era, or double-precision scientific computing, the B300 is a step backwards and a B200 contract is the right call. If your workload is four-bit inference, it is the other way round.

What 288 GB removes from your architecture.

Each of these is a job that stops needing a tensor-parallel split. Weight budgets are 1 byte per parameter at FP8 and 0.5 at FP4 or INT4, before cache.

Llama 3.1 405B on a single GPU
INT4 weights, no parallel group
~200 GB of 288

No all-reduce per layer, no parallel degree to tune, and seven cards still free for replicas or for whatever sits in front of the model.

Read the guide →
DeepSeek-V3 671B at four bits
Blackwell Ultra FP4 path, two cards
~336 GB of weights

Dense FP4 is the one axis where Blackwell Ultra pulls ahead of Blackwell, and a large mixture of experts is where that axis is loaded hardest.

Read the guide →
A programme, not a run
Sharded training held across the whole term
720 days of tenancy

The contract length is the feature here. Nothing gets reclaimed between runs, so the cluster you tuned in month three is the cluster you still have in month twenty.

Read the guide →

NVIDIA's own eight-GPU numbers.

Transcribed from NVIDIA's HGX platform specifications, both columns as printed. Two rows go the wrong way, and they are the reason this card is not automatically the better one.

HGX B300
HGX B200
Form factor
8× Blackwell Ultra SXM
8× Blackwell SXM
Total GPU memory
2.1 TB
1.4 TB
FP4 tensor core (sparse | dense)
144 | 108 PFLOPS
144 | 72 PFLOPS
FP8 / FP6 tensor core
72 PFLOPS
72 PFLOPS
FP16 / BF16 tensor core
36 PFLOPS
36 PFLOPS
TF32 tensor core
18 PFLOPS
18 PFLOPS
INT8 tensor core
3 POPS
72 POPS
FP64 / FP64 tensor core
10 TFLOPS
296 TFLOPS
NVLink, GPU to GPU
1.8 TB/s
1.8 TB/s
Networking bandwidth
1.6 TB/s
0.8 TB/s

The 2.1 TB system figure is below eight times the 288 GB part specification, so treat per-GPU capacity as something the contract states rather than something to multiply. If the INT8 or FP64 rows rule this card out for you, the B200 contract starts at 30 days; if capacity per dollar is what you are optimising, the H200 is cheaper per gigabyte than either.

Two years is long enough to do this properly.

Guides for a cluster you keep rather than one you rent for an afternoon: pipelines, batch jobs and the training stacks that assume persistent storage.

Language Models
DeepSeek-V3
Serving 671B of routed experts on two cards instead of a whole node.
Language Models
vLLM serving
Running one replica per card once the model stops needing to be split.
Training
DeepSpeed multi-GPU training
Checkpoint strategy for runs that outlive any single job submission.
Training
Hugging Face Transformers
The library layer underneath most of what a long-lived cluster runs.
Advanced
Batch processing
Keeping owned hardware busy between the jobs you actually bought it for.
Language Models
Llama 3.3 on CLORE.AI
A 70B baseline to measure the frontier model against, on the same pod.
Advanced
Multi-GPU setup
Laying out eight cards when not every job wants all eight.
See all guides →

Shorter terms, smaller cards.

B200
180 GB · Blackwell from 30 days
Compare →
H200
141 GB · cheapest per gigabyte held
Compare →
H100 80GB
80 GB · the cheapest contract of the four
Compare →

Two years of
288 GB a card.

Set the pod size and the region above and Clore returns contract terms. If two years is longer than your horizon, the B200 and H200 contracts start at thirty days.