Details
DLSS uses neural networks trained on NVIDIA's supercomputers to take lower-resolution rendered frames and game motion data as input, then output higher-quality, higher-framerate images in real time. DLSS 4, launched in January 2025 alongside the GeForce RTX 50 Series, introduced Multi Frame Generation using transformer-based AI models; on RTX 50 Series GPUs it can generate up to three additional frames per natively rendered frame (a 4X multiplier). DLSS 4.5, launched at CES 2026, introduced a second-generation transformer model for Super Resolution and a 6X Multi Frame Generation mode that generates up to five additional frames per natively rendered frame, as well as Dynamic Multi Frame Generation, which adjusts the frame multiplier automatically to match a target frame rate.
Have evidence about NVIDIA's AI practices? Submit a report.
Report a Sighting →