COMPARE / L4 VS L40S

L4 vs L40S

Specs from the chip makers' datasheets; prices from the live Compute Cafe panel, refreshed hourly. Every rate is an on-demand published price with a source receipt.

CHEAPEST L4
$0.080/hr
SpotGPUs · median $0.70 · 26 providers
CHEAPEST L40S
$0.250/hr
SpotGPUs · median $1.39 · 48 providers
CHEAPER RIGHT NOW
L4
live panel, 2026-09-27
L4L40S
MAKERNVIDIANVIDIA
ARCHITECTUREAda LovelaceAda Lovelace
LAUNCHED20232023
MEMORY24 GB GDDR648 GB GDDR6 ECC
MEMORY BANDWIDTH300 GB/s864 GB/s
TENSOR COMPUTE121 TFLOPS FP16 tensor dense (242 sparse); 242 TOPS INT8 dense362 TFLOPS FP16/BF16 tensor dense (733 sparse); 733 TFLOPS FP8 dense
POWER DRAW72 W350 W
INTERCONNECTPCIe 4.0 x16PCIe 4.0 x16 (no NVLink)

HOW TO CHOOSE

L4 vs L40S
Triple the compute and double the memory at 5x the power and price. L4 for scale-out small models, L40S for bigger single models.
Workload fit
L4: small-model serving, video pipelines, edge and burst inference. L40S: llm inference up to ~34b, fine-tuning 7b-13b, image and video generation.

L4 VS L40S - FAQ

Which is cheaper to rent, L4 or L40S?
By today's published on-demand rates: L4 starts at $0.080/GPU/hr (median $0.70, 26 providers) and L40S starts at $0.250/GPU/hr (median $1.39, 48 providers). The cheapest live listing is the L4 at SpotGPUs. Figures recompute from the panel on every refresh.
L4 or L40S - which should I rent?
L4: The L4 is the modern successor to the T4: a 72 W single-slot card with roughly double its predecessor's throughput, FP8 support, and Ada's AV1 video engines. L40S: The L40S is the workhorse of the inference-and-fine-tuning middle: Ada tensor cores with FP8, 48 GB of memory, and a price well below the H100 line. The side-by-side spec table and live price rows above give the full picture.
Where can I rent L4 and L40S?
26 providers publish L4 rates and 48 publish L40S rates on Compute Cafe. Every row links to the provider's own published price - committed-use and reserved rates are excluded.

Full pricing: L4 · L40S · all comparisons