Data updated 2026-11-04 15:09 UTC · refreshed every 5 min
Current Price
$1.21
+0.0% (24h)
Available Now
28 GPU
Good availability
Total in Network
172 GPU
Live data
Utilization
83.7%
High
Market history
USD/hour (minimum · median)
Ready to rent
Infrastructure Growth
Estimated with our VRAM calculator for a single card: 8K context, 4 concurrent users, vLLM-style serving.
35B MoE · 3B active
~77 tok/s per stream
Where the memory goes
14B
~22 tok/s per stream
Where the memory goes
4B
~64 tok/s per stream
Where the memory goes
Estimates, not benchmarks: real usage depends on the runtime, context and batch settings.
Memory
48GB GDDR6
FP32 Performance
91.6 TFLOPS
Tensor Performance
733 TFLOPS
Memory Bandwidth
864 GB/s
Architecture
Ada Lovelace
TDP
350W
FP64 Performance
1.5 TFLOPS
CUDA Cores
18176
Tensor Cores
568
The NVIDIA L40S combines AI acceleration with professional graphics capabilities for diverse workloads.
The NVIDIA L40S features 48GB of GDDR6 memory with 864 GB/s bandwidth, making it excellent for ai inference, video processing, graphics rendering. With 733 TFLOPS tensor performance and 91.6 TFLOPS FP32 compute, this NVIDIA Ada Lovelace GPU delivers exceptional performance for demanding workloads.
Currently, 172 GPU L40S GPUs are available in the GPUniq network with 83.7% utilization. The average rental price is $1.21 per hour, with 28 GPU ready for immediate deployment.
Prices are live minimum / median USD per hour on the GPUniq marketplace, as of 2026-11-04 15:09 UTC.
The current rental price for NVIDIA L40S is $1.21 per hour on GPUniq. Prices may vary based on demand and availability.
NVIDIA L40S is ideal for AI Inference, Video Processing, Graphics Rendering, Generative AI, LLM Inference. With 48GB of GDDR6 memory and 733 TFLOPS tensor performance, it excels at demanding AI and compute workloads.
Currently, there are 28 GPU available for immediate rental out of 172 GPU total in the GPUniq network. The current utilization is 83.7%.
On a single NVIDIA L40S, our VRAM calculator fits Qwen3.6 35B A3B (INT4 / AWQ, 23.8 of 48 GB); Qwen3-14B (FP16, 35.6 of 48 GB); Qwen3.5-4B (FP16, 11.4 of 48 GB) — estimated at 8K context with 4 concurrent users.
NVIDIA L40S features 48GB GDDR6 memory, 91.6 TFLOPS FP32 performance, 733 TFLOPS tensor performance, 864 GB/s memory bandwidth, and 350W TDP. It uses the Ada Lovelace architecture.
NVIDIA official specificationsAPI docsAbout GPUniqContact supportPrivacy policyTerms of use
$1.21/hr
L40S · 28 GPU