

NVIDIA GeForce RTX 4090 wins 8 of 12 benchmarks, averaging 294.4% 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 GeForce RTX 4090 leads by 640% (2.2 vs 16.28 images/min); the closest fight is Qwen3 4B (120% apart); VRAM decides part of this one: NVIDIA GeForce RTX 4090 runs 8 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 GeForce RTX 4090 | Difference |
|---|---|---|---|
| Qwen3 4B tok/s | 118.21 | 260.54 | -55% |
| Llama 3.1 8B tok/s | 76.93 | 171.29 | -55% |
| Qwen2.5-Coder 14B tok/s | 0 | 95.12 | n/a |
| Qwen3 32B tok/s | 0 | 44.28 | n/a |
| Llama 3.3 70B tok/s | 0 | 0 | n/a |
| Stable Diffusion XL images/min | 2.2 | 16.28 | -86% |
| Z-Image Turbo images/min | 0 | 7.425 | n/a |
| FLUX.1 dev images/min | 0 | 0 | n/a |
| FLUX.1 Kontext dev images/min | 0 | 0 | n/a |
| Qwen-Image-Edit images/min | 0 | 0 | n/a |
| LTX-Video (distilled) frames/s | 0 | 4.7 | n/a |
| Wan 2.2 5B (720p) frames/s | 0 | 0.43 | n/a |
| NVIDIA GeForce RTX 3060 Ti | NVIDIA GeForce RTX 4090 | |
|---|---|---|
| VRAM | 8GB | 24GB |
| Transistors | 17,400M | 76,300M |
| Die size | 392.5 mm² | 608.5 mm² |
| Transistor density | 44.3 M/mm² | 125.4 M/mm² |
| Architecture | Ampere (GA104) | Ada Lovelace |
| Memory bandwidth | 448 GB/s | 1008 GB/s |
| Boost clock | 1,665 MHz | 2,520 MHz |
| TDP | 200 W | 450 W |
| Launch MSRP | $399 | $1,599 |
| Release | 2020-12-02 | 2022-10-12 |
NVIDIA GeForce RTX 3060 Ti full review · NVIDIA GeForce RTX 4090 full review · All AI & Machine Learning rankings