GPU Cloud
L40S, dedicated.
A single-GPU L40S machine, 48 GB of GDDR6 on NVIDIA's Ada Lovelace architecture, on an x86 host. $0.99 per hour, billed per minute, the same terms as every other GPUwerk tier.
Rented first come, first served: deploy an L40S when capacity is free, or join its waitlist in the console.
Specifications
Per node. Every GPUwerk instance, on any tier, is one whole machine to yourself.
| GPU | L40S, Ada Lovelace architecture |
|---|---|
| Memory | 48 GB GDDR6 |
| Host | x86 host |
| Tenancy | One customer per machine, never shared |
GPUwerk publishes tok/s and training throughput figures only for runs on its own hardware. See the DGX Spark fleet benchmarks for how they are reported.
What would you run on it?
48 GB positions the L40S between the 5090's 32 GB and a Spark's 128 GB, with data-center-class ECC memory.
Mid-size model inference
Models up to about 40B parameters at FP8, or a 30B-class model with room for the KV cache.
Mixed graphics and AI
RT cores and a media engine next to FP8 tensor cores, for pipelines that render and infer on one card.
LoRA fine-tunes
Adapter training on 7B to 13B-class models, which fits in 48 GB with headroom.
Poor fit: multi-gpu training
It has no NVLink, so a job that needs several GPUs exchanging data quickly belongs on an 8-GPU server.
What fits in 48 GB
Weights only: memory divided by bytes per parameter. Leave headroom for the KV cache, activations and the runtime, often 10 to 30% more.
| Precision | Per parameter | Largest model, weights only |
|---|---|---|
| FP16 / BF16 | 2 bytes | about 24B parameters |
| FP8 | 1 byte | about 48B parameters |
| 4-bit quantized | 0.5 bytes | about 96B parameters (no native FP4 on Ada; the runtime dequantizes) |
Billing
| Rate | $0.99 per hour, billed per minute |
|---|---|
| Billing | Per minute, against a prepaid credit balance |
| Fees | No egress fees |
| Commitment | None |
Prices are in USD, excluding applicable tax. Full billing mechanics, including what a stop versus a terminate does, are on the pricing page.
FAQ
Can I rent an L40S now?
Deploy it from the console when capacity is free; while none is, the console offers a waitlist and people in line are emailed one at a time, oldest first. Capacity is not held for anyone: whoever deploys first gets it.
How does billing work on this tier?
The same as every GPUwerk tier: per minute, against a prepaid credit balance, in USD, excluding applicable tax, with no egress fees and no commitment.
Need GPU time today?
A DGX Spark has 128 GB of unified memory and is rentable now, at $0.79/hour billed per minute. First top-up: pay $20, get $30 in credit. See how a Spark compares with a discrete card, or join the L40S waitlist.