NVIDIA B300 — what a GPU-hour costs
Listed rates, not availability. 24 rates read between and . · Renting for a year or more? Reserved quotes ↓
In September 2026 the B300 like-for-like median did not move.
Size your workload
Who is cheapest today, and what a month costs.
GPUsHours per monthRegionB300 by provider
One table per market, on-demand first; cheapest first within each.
| Provider · what is sold | $ / GPU-hour | 30 days | Read | Source |
|---|---|---|---|---|
 TensorPoolGPU only | $5.49−39% vs median | stoppedlast confirmed · not in the figures | ||
 Deep Infra7d +20.4% | $5.89−34% vs median | +20.4% · 26 d | captured | Deploy ↗ |
 DaytonaGPU only | $6.25−30% vs median | flat · 27 d | captured | Deploy ↗ |
 Gcore$53.78/node ÷ 8 · Bare Metal GPU Clusters · €6.00 listed · ECB 1.1205 on October 22, 2026 | $6.72−25% vs median | flat · 26 d | captured | Deploy ↗ |
 FPT AI FactorySoutheast Asia | $6.99−22% vs median | flat · 11 d | captured | Deploy ↗ |
 Hyperstack | $7.40−18% vs median | flat | captured | Deploy ↗ |
 Civo$59.92/node ÷ 8 | $7.49−17% vs median | flat · 22 d | captured | Deploy ↗ |
 Enverge | $7.50−17% vs median | flat · 27 d | captured | Deploy ↗ |
 Scaleway$67.16/node ÷ 8 · fr-par-2 · €7.50 listed · ECB 1.1193 on October 23, 2026 | $8.39−7% vs median | flat | captured | Deploy ↗ |
 Sestercefrom | $8.70−3% vs median | +75.8% | captured | Deploy ↗ |
 Crusoe | $8.96−0% vs median | −9.5% · 1 d | captured | Deploy ↗ |
 Barrack AI | $8.990% vs median | flat · 10 d | captured | Deploy ↗ |
 GPU.aifrom | $8.990% vs median | +13.9% · 22 d | captured | Deploy ↗ |
 RunPodAPISecure Cloud · the API request | $8.990% vs median | +13.9% | captured | Deploy ↗ |
 Nebius | $9.50+6% vs median | flat | captured | Deploy ↗ |
 Together AI | $9.99+11% vs median | flat · 25 d | captured | Deploy ↗ |
 VerdaAPI7d +7.1% | $10.69+19% vs median | +31.7% | captured | Deploy ↗ |
 Lyceum | $11.09+23% vs median | +38.8% · 27 d | captured | Deploy ↗ |
 PaperspaceGPU Droplets | $11.19+24% vs median | flat | captured | Deploy ↗ |
 Fireworks AIGPU only | $15.00+67% vs median | flat · 21 d | captured | Deploy ↗ |
 Oracle Cloud (OCI)API | $15.00+67% vs median | flat | captured | Deploy ↗ |
 EmpirioLabs7d −30.4% | $16.00+78% vs median | −3.0% · 27 d | captured | Deploy ↗ |
 Amazon Web ServicesAPI$142.42/node ÷ 8 · us-east-1 | $17.80+98% vs median | flat · 28 d | captured | Deploy ↗ |
| Provider · what is sold | $ / GPU-hour | 30 days | Read | Source |
|---|---|---|---|---|
| Nebiusfrom | $0.99−89% vs median | unconfirmedlast confirmed · not in the figures | Deploy ↗ | |
| DaytonaGPU only−$2.17 (−35%) vs its on-demand $6.25 | $4.08−55% vs median | flat · 27 d | captured | Deploy ↗ |
 CoreWeave$35.84/node ÷ 8 · North America | $4.48−50% vs median | flat | captured | Deploy ↗ |
| CoreWeave$36.70/node ÷ 8 · Europe | $4.59−49% vs median | flat | captured | Deploy ↗ |
| VerdaAPI−$5.34 (−50%) vs its on-demand $10.69 · 7d +7.1% | $5.34−41% vs median | +31.7% | captured | Deploy ↗ |
| Lyceum−$5.50 (−50%) vs its on-demand $11.09 | $5.59−38% vs median | +40.1% · 16 d | captured | Deploy ↗ |
| PaperspaceGPU Droplets · −$5.02 (−45%) vs its on-demand $11.19 | $6.17−31% vs median | −22.9% | captured | Deploy ↗ |
| Provider · what is sold | $ / GPU-hour | 30 days | Read | Source |
|---|---|---|---|---|
| RunPodAPICommunity Cloud · the API request | $6.94−23% vs median | flat | captured | Deploy ↗ |
 LiumfloorAPI$64.00/node ÷ 8 · Multi-GPU machines | $8.00−11% vs median | thin marketfewer offers than a floor needs · last floor · not in the figures | Deploy ↗ | |
| LiumfloorAPI | $8.00−11% vs median | flat · 6 d | captured | Deploy ↗ |
| Provider · what is sold | $ / GPU-hour | 30 days | Read | Source |
|---|---|---|---|---|
 ModalGPU only | $7.10−21% vs median | flat | captured | Deploy ↗ |
 fal | $12.99+44% vs median | +52.8% | captured | Deploy ↗ |
| Provider · what is sold | $ / GPU-hour | 30 days | Read | Source |
|---|---|---|---|---|
 Phala Cloudmin. commitment 1mo · 7d +30.8% | $8.50−5% vs median | +30.8% · 12 d | captured | Reserve ↗ |
| Provider · what is sold | $ / GPU-hour | 30 days | Read | Source |
|---|---|---|---|---|
| Civo$51.92/node ÷ 8 · −$1.00 (−13%) vs its on-demand $7.49 | $6.49−28% vs median | flat · 22 d | captured | Reserve ↗ |
| PaperspaceGPU Droplets · −$3.25 (−29%) vs its on-demand $11.19 | $7.94−12% vs median | flat | captured | Reserve ↗ |
| Provider · what is sold | $ / GPU-hour | 30 days | Read | Source |
|---|---|---|---|---|
 TheAI Cloud$37.60/node ÷ 8 | $4.70−48% vs median | captured | Reserve ↗ | |
| Civo$43.92/node ÷ 8 · −$2.00 (−27%) vs its on-demand $7.49 | $5.49−39% vs median | flat · 22 d | captured | Reserve ↗ |
Other cards today, median rate card per GPU-hour: H100 SXM $3.53 (Price Index) · B200 $7.50 · H200 $4.56 · GB300 NVL72 $18.00 · MI325X $2.25 · RTX 4090 $0.49 · all cards
Renting the B300 for a year or more?
The cheapest committed B300 rate each provider publishes, by term. A commitment is paid whether the GPU runs or not.
| Term | Per GPU-hour | Provider | vs its on-demand |
|---|---|---|---|
| 1-year | $6.49 read | Civo cheapest of 2 providers | −13% ($7.49) |
| 3-year | $4.70 read | TheAI Cloud cheapest of 2 providers | — |
| Other commitment 1mo | $8.50 read | Phala Cloud | — |
Across every card, a 1-year rate runs a median −15% against the same provider’s on-demand rate (7 providers); a 3-year rate −30% (5 providers). Every committed rate, card by card.
Get quotes for the B300. A person at GPU Economy reads your request and passes it to up to three providers that list a rate for the card, the cheapest listed first; the quotes come from them, directly to you.
We may be paid a fee if a request becomes a contract; who receives it follows the listed rates, never a fee, and it changes no published rate or the Price Index (how, privacy).
As of , the cheapest listed rate for the NVIDIA B300 is $4.08 per GPU-hour on Daytona (spot). The cheapest on-demand rate is $5.89 on Deep Infra. Across 24 providers, the median rate card — each provider’s cheapest on-demand rate, community capacity and marketplace floors out — is $8.99 per GPU-hour, and the middle half of those rate cards sits between $7.47 and $11.12. Between rate cards, on-demand rates for the same card run from $5.89 (Deep Infra) to $17.80 (Amazon Web Services), 3.0× apart across 24 providers. By market, the cheapest listed rates are $4.08 spot (Daytona, GPU only), $6.94 on community capacity (RunPod), $5.89 on-demand (Deep Infra), $8.50 reserved with no stated term (Phala Cloud), $6.49 on a 1-year commitment (Civo) and $4.70 on a 3-year commitment (TheAI Cloud). On a single rate card, Civo lists it at $7.49 on-demand, $6.49 on a 1-year commitment and $5.49 on a 3-year commitment: a commitment is paid whether the GPU runs or not. With 288 GB of VRAM, that is $0.020 per GB of memory per hour at the cheapest on-demand rate and $0.031 at the median.
The two cheapest on-demand rate cards for the NVIDIA B300, side by side on every card they both list: Deep Infra vs Daytona.
Dated pieces on the NVIDIA B300: — Lyceum H100 went $3.49 to $4.69 an hour, B300 $11.09, October 16, 2026 · — B300 rents from $4.95 to $17.80 a GPU-hour, September 2026. Each is frozen as of its date.
As of , 4 providers price the NVIDIA B300 on-demand as a whole 8-GPU node (HGX, as CoreWeave, Crusoe, Nebius and Paperspace name it in their own listings): from $53.78 per node-hour (Gcore, 8×B300 SXM, Bare Metal GPU Clusters, $6.72 per GPU) to $142.42 (Amazon Web Services, p6-b300.48xlarge, us-east-1, $17.80 per GPU), 2.6× apart. 8 GPUs at the cheapest on-demand rate ($5.89 on Deep Infra) come to $47.12 an hour, where a provider rents 8 to one account and they need not share a unit. A unit price is what the provider quotes for the 8 GPUs together; whether it also rents one is not part of the quote.
How it is computed
What we measure
Publicly listed rental prices for 79 GPU models (NVIDIA and AMD) at 99 providers — 1165 current quotes across seven markets — and the retail price of the cards.
Where it comes from
Each provider's own pricing page or API; retail from reseller pages. Every figure links back to its source.
How it is normalised
Node prices are divided by GPU count, per-second rates multiplied out, a price listed in another currency shown as listed, beside its USD figure at the dated official rate (ECB; the Bank of Russia for RUB and VND). A reservation published as a total for its term is divided by the hours in that term, 8,760 per year.
What is modelled, not measured
One figure: cost per million output tokens, a listed rate divided by the throughput NVIDIA published for that card, published with the run that produced it in /token-cost.json, not on this page. Risk is not a model — it is read off the market a quote is from.
What we never do
Never estimate a price, never convert a currency silently, never publish a number a reader cannot verify on the linked page.
1165 quotes · 99 providers. Seven files under CC BY 4.0; history is stored append-only and never rewritten.
Weekly history since 2025, with a like-for-like table per provider: NVIDIA B300 price history.
Retail listings
No retail listings tracked for this card in the last 30 days.
Compare the NVIDIA B300 with: NVIDIA B200
Every rate on this page as data: /gpu/b300/prices.json — the same fields as prices.json, narrowed to this card. CC BY 4.0, no key, no rate limit. On your own page: embed this card’s cheapest rates (/embed/gpu/b300, an iframe, no script).
Data updated: