NVIDIA’s latest open event was GDC 2026, and they brought forward several pieces of news for both GeForce-related gaming announcements as well as the “non-GeForce” stuff – basically all the AI software.
First off the list is the official launch of DLSS 4.5 featuring Dynamic Multi Frame Generation and 6x Multi Frame Generation usable by fellow GeForce RTX 50 series GPU owners for absurd levels of frame rate or an ever-changing MFG tech that adjusts accordingly to performance and other metrics.
Players will be able to turn on these in supported games starting March 31.
Meanwhile, a couple more games have been revealed to implement path tracing and/or DLSS 4.5 for more visual fidelity:
- 007 First Light
- Aniimo
- Barkour
- CONTROL Resonant
- Cthulhu: The Cosmic Abyss
- Directive 8020
- Edge of Memories
- Endurance Motorsport Series
- Gray Zone Warfare
- INDUSTRIA 2
- Samson
- Sea of Remnants
- StarRupture
- STAR WARS: Galactic Racer
- The Mound: Omen of Cthulhu
- The Vernyhorn
- Tides of Annihilation
- WARDOGS
- War Thunder
- Where Winds Meet
Next up on the list is the NVIDIA RTX Mega Geometry Foliage that’s built from the foundations of RTX Mega Geometry, with both technologies sharing the advantage of drastically reducing VRAM usage while improving frame rates in cases like open world where trees, fields, and whatnot are filled to the brim for maximum immersion.
This new system will first be adapted into the upcoming The Witcher 4.
On the other hand, the community modding tool RTX Remix will be getting Advanced Particle VFX that will uplift everything that is visual effects-related, on top of the existing functions the suite provides, such as 4K path-traced visuals, enhanced PBR assets, RTX Remix Logic, and more.
Currently, they are demonstrating with a Quake III Arena RTX Mod map, and soon modders will be able to use it across supporting titles.
As part of the announcement, some games also received DLSS updates and/or will be launched with them, and these are:
- DEATH STRANDING 2: ON THE BEACH: DLSS 4 with Multi Frame Generation (RTX 50 Series); DLSS Super Resolution; DLAA; DLSS 4.5 Super Resolution via NVIDIA app overrides; Ultrawide support; 4K enhanced visuals.
- Marathon: DLSS Super Resolution; DLAA; DLSS 4.5 Super Resolution via NVIDIA app overrides; NVIDIA Reflex.
- Monster Hunter Stories 3: Twisted Reflection: DLSS Super Resolution; DLAA; NVIDIA Reflex.
- Demonologist: DLSS 4.5 Super Resolution; DLSS Frame Generation; NVIDIA Reflex; DLSS 4 with Multi Frame Generation via NVIDIA app (RTX 50 Series).
- Black One Blood Brothers: DLSS Super Resolution; DLSS 4.5 Super Resolution via NVIDIA app overrides.
A new version of the GeForce Game Ready Driver has been released under 595.79, and you can grab it here or via the NVIDIA App.
For the cloud gaming fans, from March 19 onwards, GeForce NOW will support up to 90FPS gameplay from the previous 60FPS max on virtual reality devices, including the Apple Vision Pro and Meta Quest, and NVIDIA will also be optimizing what games can be run exactly for each user via GeForce NOW without repurchasing titles. n-app labels will show members which games are already playable on GeForce NOW through their connected Ubisoft Connect and Xbox Game Pass subscriptions, while support for new Gaijin sign-on, plus GOG sign-on and library linking, joins existing support for Xbox, Ubisoft Connect, and Epic.
In the coming months, Team Green will be releasing NVIDIA CloudXR 6.0 for visionOS that allows Apple Vision Pro users to stream immersive PC and cloud experiences from NVIDIA RTX systems, with iRacing and X-Plane 12 being some of the very first titles to support this feature.
It is enabled by a dedicated streaming platform that delivers high-fidelity, low-latency XR from a local RTX PC or cloud GPU. CloudXR 6.0 adds visionOS integration with privacy-preserving foveated streaming, which helps developers deliver better visual quality and performance without rebuilding content for standalone hardware. For developers, this means faster iteration, easier deployment of demanding spatial workloads, and a practical path to bring PC-class experiences to Apple Vision Pro.
Coming in a future update, Microsoft’s DirectX API will be supporting AI from the get-go as NVIDIA wants to standardize hardware-accelerated AI in games for improved GPU efficiency, reduce context switching, and eliminate common issues like shader compilation stutter.
This is made from a summer’s preview of Cooperative Vector support, which already allowed developers to accelerate certain AI workloads – such as neural texture compression – using Tensor Cores. Now, Microsoft plans to push things further by introducing two new technologies: DirectX Linear Algebra and DirectX Compute Graph Compiler. These are designed to help accelerate the entire gaming pipeline by making AI workloads more efficient and better integrated with graphics rendering.
Additionally, the use of Microsoft Advanced Shader Delivery, or ASD, shaders is precompiled and distributed directly during the game download process to allow users to solve real-time shader compilation issues.
Lastly, the popular generative AI tool ComfyUI now offers a new App View dedicated to those who aren’t so tech-savvy in the world of setting up gen AI stuff. No complex node graphs – Just a few prompts, tweak a few parameters, then the results will be in front of you in just a moment.
More optimized AI models are also joining the lineup. New versions such as FLUX.2 Klein 4B and FLUX.2 Klein 9B are already available, while support for LTX-2.3 with NVFP4 is expected soon. Creators can download the model checkpoints from Hugging Face, load the workflows in ComfyUI’s Template Browser, and swap in the optimized models.
NVIDIA is also tackling another pain point in AI video creation – upscaling. Artists often generate lower-resolution previews first because producing full 4K outputs can be slow and memory-intensive. To address that, NVIDIA is bringing RTX Video Super Resolution directly into ComfyUI as a workflow node.
Powered by the NVIDIA Video Effects SDK and running on RTX Tensor Cores, the technology can upscale generated videos to 4K up to 30x faster than many local upscaling tools, while also consuming significantly less VRAM. NVIDIA is also making the technology accessible to developers through a Python package available on PyPI, along with sample code on GitHub to help integrate the feature into custom AI pipelines.
