NVIDIA L4
The L4 is a low-profile, energy-efficient inference accelerator with 24GB GDDR6 drawing just 72W. Ideal for high-density inference deployments in space and power-constrained environments.
VRAM
24 GB
Memory
GDDR6
Bandwidth
300 GB/s
FP16
121 TF
TDP
72 W
Memory and compute
Architecture and power
Notes
Low-power inference
Models that may fit on NVIDIA L4
Estimates based on INT8 quantization. Actual fit depends on framework and batch size.
More GPUs by NVIDIA
The manufacturer's other cards in the catalog, largest memory first, with bandwidth and power where published.
Frequently asked questions about NVIDIA L4
Answered from the entry's own figures: memory, which models fit, bandwidth, power and the card's class.
NVIDIA L4 has 24 GB of GDDR6 at 300 GB/s. At INT8 a model's weights take about one byte per parameter, so the card holds roughly a 19B-parameter model with room for context; larger models need several cards or a lower precision.
From the catalog's estimates at INT8, NVIDIA Nemotron Nano 12B V2 VL BF16, Qwen 2.5 14B Instruct, Gemma 3 12B and Moonlight 16B A3B Instruct fit on a single NVIDIA L4 with their default context. The list on this page shows the estimated memory each takes; the capacity planner sizes any model against this card for your own context length, batch size and request rate.
300 GB/s. Bandwidth bounds how fast a model generates tokens, because every token reads the whole set of weights and the KV cache from memory; at this rate a 19B model at INT8 could be read on the order of 16 times a second on one card, before batching and framework efficiency.
NVIDIA L4 has a TDP of 72 W, with a maximum of 72 W, in the PCIe Low-profile form factor. Budget the TDP per card plus host overhead when sizing a server's power and cooling.
NVIDIA L4 is a datacenter accelerator, meant for servers and multi-GPU serving, built on the Ada Lovelace architecture and released in 2023. It has no multi-GPU interconnect, so a model must fit on one card or be split over the slower PCIe bus.
Specifications from the manufacturer's published material; a figure that is not published is shown as such, and a derived one is explained in the notes.
Added Jan 25, 2026 · Last updated Jan 25, 2026
Explore more
Companion tools that draw on the same catalog and routing engine.
Put intelligence in the middle.
One gateway in front of every model, with your policies applied and every decision on record. Start on the free plan with 5,000 requests a month, no card required.