

NVIDIA L40S wins 12 of 12 benchmarks, averaging 276.4% faster.
Both cards were measured first-party on our bench, same suite, same test rig.
The gap is widest in Stable Diffusion XL, where NVIDIA L40S leads by 674% (2.2 vs 17.02 images/min); the closest fight is Llama 3.1 8B (76% apart); VRAM decides part of this one: NVIDIA L40S runs 152 of our 12 AI workloads while the other card runs 3, models that don't fit score zero.
| Benchmark | NVIDIA GeForce RTX 3060 Ti | NVIDIA L40S | Difference |
|---|---|---|---|
| Qwen3 4B tok/s | 118.21 | 212.05 | -44% |
| Llama 3.1 8B tok/s | 76.93 | 135.51 | -43% |
| Qwen2.5-Coder 14B tok/s | 0 | 74.6 | n/a |
| Qwen3 32B tok/s | 0 | 34.44 | n/a |
| Llama 3.3 70B tok/s | 0 | 16.49 | n/a |
| Stable Diffusion XL images/min | 2.2 | 17.02 | -87% |
| Z-Image Turbo images/min | 0 | 8.325 | n/a |
| FLUX.1 dev images/min | 0 | 3.836 | n/a |
| FLUX.1 Kontext dev images/min | 0 | 1.693 | n/a |
| Qwen-Image-Edit images/min | 0 | 1.06 | n/a |
| LTX-Video (distilled) frames/s | 0 | 8.01 | n/a |
| Wan 2.2 5B (720p) frames/s | 0 | 0.49 | n/a |
| NVIDIA GeForce RTX 3060 Ti | NVIDIA L40S | |
|---|---|---|
| VRAM | 8GB | 48GB |
| Architecture | Ampere (GA104) | Ada Lovelace |
| Memory bandwidth | 448 GB/s | 864 GB/s |
| Boost clock | 1,665 MHz | 2,520 MHz |
| TDP | 200 W | 350 W |
| Launch MSRP | $399 | $7,500 |
| Release | 2020-12-02 | 2023-08-08 |
NVIDIA GeForce RTX 3060 Ti full review · NVIDIA L40S full review · All AI & Machine Learning rankings