L40S GPU Rental Prices & Live Availability
As of , the cheapest single L40S confirmed in stock is $0.38 per GPU-hour on Lium. GPU Finder compares 12 providers for this size and refreshes availability hourly. Methodology.
Structured offer basis: 89 current 1× L40S listings range from $0.38 to $10.99 USD per physical GPU-hour. Whole-node prices for other GPU counts are not mixed into this range.
Cheapest single L40S rental
verifiedCheapest L40S confirmed in stock right now
Limited- Is now a good time to rent an L40S cloud GPU?
- ▼ near its 6-month low — The 4× L40S on-demand floor is $0.92/hr ($0.23/GPU-hr) — 6-month range $0.92–$3.40/hr for 4× nodes. In short, a relatively good time to rent — the floor price is near its 6-month low.
- What is the cheapest L40S cloud GPU available right now?
- Cheapest L40S confirmed in stock right now: $0.38/GPU-hr on Lium (1× listing, $0.38/hr).
- Which cloud providers most reliably have the L40S in stock?
- Most reliably in stock over the last 30 days: PrimeIntellect, AceCloud, Shadeform. Frequently waitlisted: Vast, Cudo.
Prices are the cheapest listed on-demand rates; “in stock” means the provider reported capacity on our last hourly check. How we compute this →
L40S rental prices by provider
- datacrunch__1xL4…_48GB__20__60In stock83% · 30d$0.91/hr1× L40S20 vCPU60 GBFree egressFI
Get notified when L40S drops in price or reliability changes
One email per change, max once a day. We send a confirmation link first; one-click unsubscribe in every email.
L40S (1× GPU) availability — last 7 days
2 of 10 providers kept 1× L40S ≥80% available all week. Polled hourly · 10 providers tracked · 7-day window.
| Provider | Fri11 | Sat12 | Sun13 | Mon14 | Tue15 | Wed16 | Thu17 | Week |
|---|---|---|---|---|---|---|---|---|
| AceCloud | 100% | |||||||
| PrimeIntellect | 100% | |||||||
| Nebius | 92% | |||||||
| Lium | 74% | |||||||
| Packet | 72% | |||||||
| Shadeform | 69% | |||||||
| Runpod | 21% | |||||||
| Cudo | 0% | |||||||
| Vast | 0% | |||||||
| Vultr | 0% |
30-day reliability depth
GPU Finder does not treat availability as a static yes/no flag. For L40S, current stock badges are paired with a 30-day reliability score based on our hourly stock checks — the share of tracked time each listing was reported available. How we compute this →
- 88
- listings with a 30-day score
- 45%
- of listings on this page have a score
- 30d
- longest stock history on this page
Strongest reliability signals here: PrimeIntellect (100%, 30d covered), AceCloud (99%, 30d covered), Shadeform (94%, 30d covered).
Caveat: scores stay hidden until a listing has at least 48 tracked hours, and some provider APIs expose coarse capacity levels instead of exact stock counts. Use the score as historical depth beside current availability, not as a guarantee that a GPU will still be allocatable when you click through.
Price History & Comparison
Full L40S price history (20 months) →Spot pricing decision guide
Treat spot as a risk-adjusted capacity decision, not just a cheaper number.
per GPU-hour
30% below on-demand floor
daily median for the same offers with complete history
median $1.25/GPU-hr
Cheapest trusted spot / interruptible capacity is on Vast; 2 providers currently report spot or interruptible stock (138 units/signals visible).
Nebius has kept spot or interruptible capacity in stock most reliably over the last 30 days (63%). See the availability section for the broader stock history.
The discount is visible, but availability, reliability, or volatility argues for more caution.
Good fit: checkpointed batch jobs, flexible training, and experiments that can restart. Avoid for production serving, deadline-bound runs, or jobs that cannot tolerate eviction.
Source caveat: Runpod exposes explicit spot fields; Vast is labeled interruptible/bid-floor. Lambda is treated as on-demand-only unless a verified spot field is added.
About the L40S
The NVIDIA L40S occupies the sweet spot between consumer GPUs and full datacenter accelerators. Based on Ada Lovelace, it provides 48 GB of GDDR6 with ECC and FP8 support in a standard PCIe, 350W form factor. It is purpose-built for inference-heavy deployments and workloads that need more memory than an RTX 4090 but do not justify H100-class pricing.
Key Specifications
| Architecture | Ada Lovelace (AD102) |
| GPU Memory | 48 GB GDDR6 with ECC |
| Memory Bandwidth | 864 GB/s |
| FP16 Tensor Core | 362 TFLOPS |
| TDP | 350W |
| Interconnect | PCIe Gen4 x16 |
| Release Year | 2023 |
Cloud Pricing Context
L40S on-demand pricing starts as low as $0.32/hr on marketplace providers, with mainstream options around $1.57/hr. Its PCIe form factor means more providers can offer it without specialized NVLink infrastructure, keeping prices competitive.
Best For
- Production inference serving for models in the 7B-13B range
- Video and image generation pipelines (Stable Diffusion, Flux)
- Mixed AI and visualization workloads using RT cores
- Teams that need ECC memory at mid-tier pricing
4× L40S server pricing and bundle math
Live-derived multi-GPU estimates for 4× L40S (192 GB aggregate VRAM).
Normalized from live on-demand rows by GPU count.
Cheapest live 4-GPU instance.
Approx. 24 × 30 at the bundle hourly rate.
GPU Finder currently tracks a live 4-GPU L40S instance from $1.52/hr ($1094.40/mo at 24×30). Use the table below to compare exact configurations and availability.
Networking, CPU, host memory, storage, and egress vary by provider and are not folded into this estimate. Prices are for on-demand cloud instances unless the provider table shows a spot or interruptible rate.
4× L40S server pricing FAQ
How much does a 4× L40S server cost per hour?
Based on current live rates, the cheapest 4× L40S bundle starts around $1.52/hr, which is roughly $1094.40/mo at 24×30 utilization. The query "4x nvidia l40s server price 2026" is the most common form of this search. L40S is a PCIe GPU, so 4× setups are usually workstation or small servers rather than NVLink clusters.
What is the monthly price for a 4× L40S server?
A 4× L40S server at $1.52/hr is about $1094.40/mo at 24×30. Add egress, storage, and CPU/RAM costs per provider.
How does 4× L40S compare to 4× RTX A6000?
Both 4× L40S and 4× RTX A6000 expose 48 GB of VRAM per GPU and use PCIe. L40S adds a Transformer Engine and FP8 support for inference, while RTX A6000 is a workstation card that can be cheaper for rendering and development workloads.
Where can I rent a single L40S GPU?
Use the live L40S provider table on this page. L40S is widely available from both neoclouds and mainstream cloud providers because its PCIe form factor lowers host complexity.
Is a 4× L40S server good for LLM inference?
Yes, for 7B–13B parameter models and image/video generation workloads that fit in 48 GB of GDDR6. L40S lacks NVLink, so large distributed training jobs are not its best fit, but it is one of the most cost-effective inference options with ECC memory.