Apex Neural Rendering Takes On DLSS With Beta for Older GPUs
Apex Compute has released a beta of Apex Neural Rendering's frame generation tool, which it says runs faster and uses 62% less video memory than NVIDIA DLSS. It supports GeForce GTX 16 cards and AMD Radeon RX 5000 and newer. The figures come from the company's own tests.

Apex Compute has released a beta of Apex Neural Rendering (ANR), a cross-platform frame generation tool that the company says outperforms NVIDIA's DLSS, and the Apex neural rendering challenges to NVIDIA start with hardware support. The latest build, FG V.B14, was published on November 12, 2026.
The beta works on cards that lack official DLSS (Multi) Frame Generation support. Gamers can download it from GitHub today. Only the frame generation part is available so far.
What ANR includes
Apex announced the suite in an official X thread. It has three parts: Frame Generation, Super Resolution and Ray Reconstruction. The upscaling and ray-reconstruction models are still in development.
Frame generation inserts extra, computer-made frames between rendered ones to make motion look smoother.
Supported hardware
On the NVIDIA side, ANR covers GeForce GTX 16 and RTX 20, 30, 40 and 50 Series cards. On AMD, it covers Radeon RX 5000, 6000, 7000 and 9000 Series and the Steam Deck.
The beta runs on Windows 10 and 11, and on Linux and SteamOS through Proton. Apple silicon gets a separate native Metal test kit for macOS 26 and newer.
Apex's benchmark claims
The figures are Apex's own. On a GeForce RTX 5090 at 3840×2160, ANR needs 1.01 milliseconds per generated frame in 2X mode, against 1.26 ms for NVIDIA DLSS Frame Generation. In 4X mode the numbers are 0.61 ms and 0.82 ms.
Memory is the bigger gap. ANR uses 214 MB at 4K, versus 562 MB for NVIDIA's tool, a cut of about 62%. Its neural network weights take up just 0.17 MB.
The headline "up to 2.4X faster" claim looks larger than the RTX 5090 data suggests. Those published results show more modest gains in processing time. They also measure the cost of generating extra frames, not an equivalent rise in real frames per second.
Image quality and testing
Apex's interactive comparisons point to better handling of particles, moving objects, water, vegetation and disocclusion artifacts, which appear when a hidden area of the scene is suddenly revealed. The company also published video showcases of The Last of Us Part II Remastered and Hogwarts Legacy.
Its internal CGVQM-5, ColorVideoVDP and PSNR quality measurements favor ANR over DLSS FG. Independent testing is still needed to confirm any of it.
How to try it
Installation goes through OptiScaler, a tool for swapping upscaling and frame generation in games. Apex offers separate packages for games that already have OptiScaler and those that don't. Users extract the files, run the installer in the game's executable folder, and switch on frame generation in the game's graphics settings.
The V.B14 release also fixes a Windows problem that stopped some games launching with older Visual C++ runtimes. Apex has also released nr-record, a utility that captures frame-generation inputs and replays them through competing implementations for side-by-side image comparisons.
The code is not open source yet. Apex's developers have promised to release it later. As an experimental beta, it can be expected to show compatibility problems and visual artifacts.
















