Skip to content
ShiftDelete.Net
Search news/ TR Türkçe
News Feed AI Smartphones Games Cars Computers Social Media Space Reviews Latest news
  1. Home
  2. Graphic Cards

Apex Neural Rendering Arrives as a Cross Platform DLSS Rival

Apex Compute unveiled Apex Neural Rendering on November 11, 2026. Frame Generation is in beta now, and the company's 2.4x speed claim comes from its own tests.

Hakkı ALKANHakkı ALKAN

Published 4 min read

Particle and propeller scene from an Apex Neural Rendering frame generation demo

Apex Compute announced Apex Neural Rendering (ANR) on November 11, 2026. In the company's X post, ANR is a cross-platform package that combines Super Resolution, Ray Reconstruction, and Frame Generation. Frame Generation (ANR FG) is available as a beta download, while Super Resolution and Ray Reconstruction are still in training. Apex Compute also claimed up to 2.4x faster performance and higher quality rendering than NVIDIA DLSS; those figures come from the company's own measurements.

Particle and propeller scene from an Apex Neural Rendering frame generation demo

What is Apex Neural Rendering, and how does it run?

According to the official Apex Neural Rendering page, the technology can run on NVIDIA GeForce RTX 20-series and newer plus GTX 16, AMD Radeon RX 5000-series and newer plus Steam Deck, Intel Arc A and B, and Apple silicon from M1 onward. The listed platforms are Windows 10 and 11, Linux and SteamOS with Proton, and macOS 26 or later.

The beta comes in two packages: anr-fg-full for games that do not use OptiScaler, and anr-fg-addon for games that already do. Installation notes say frame generation is turned on through the NVIDIA option the game already offers. That is not a first-party integration from the game publisher; it is a beta delivered through OptiScaler.

ANR FG packages and release notes are published in the company's GitHub repository. FG V.B13, released on November 12, 2026, also adds a test kit for Apple silicon on macOS 26 or later. The company presents that kit as a way to measure ANR FG on Metal against Apple MetalFX frame interpolation on the same frames.

What is the 2.4x claim based on?

On the official site table, Apex Compute lists 1.01 ms per generated frame for ANR FG and 1.26 ms for NVIDIA DLSS Frame Generation on a GeForce RTX 5090 at 3840 x 2160. In 4X mode, the mean time is 0.61 ms for ANR FG and 0.82 ms for DLSS FG. That gap is about 1.25x and should not be treated as the same measurement as the 2.4x claim on X.

A separate chart the company posted on X lists 2X / 3X / 4X frame generation at 4K on an RTX 5090 as 0.34 / 0.53 / 1.13 for APEX and 0.81 / 1.08 / 1.27 for NVIDIA DLSS-G. The chart does not label the unit. The 2X pair (0.34 versus 0.81) is close to the company's 2.4x headline, but it is still a company-provided chart, not an independent test. AMD's own neural rendering plans are covered in our earlier piece, AMD Develops Neural Rendering Technology to Rival NVIDIA DLSS 5.

What do the quality and memory tables show?

The official quality table also compares ANR FG with DLSS FG. In the 2X row, the CGVQM-5 mean patch score is 96.59 for ANR versus 93.66 for DLSS; ColorVideoVDP is 9.24 versus 8.77 JOD, and PSNR is 33.84 versus 31.22 dB. The company published 95% confidence intervals as well. These are not independent lab results.

A separate company table lists GPU memory use at 4K as 214 MB for ANR FG and 562 MB for DLSS FG. The network weights file is given as 0.17 MB.

ItemApex dataNote
ANR componentsSR, RR, FGSR and RR are in training; FG is in beta
Supported GPUsNVIDIA RTX 20+ and GTX 16; AMD RX 5000+ and Steam Deck; Intel Arc A/B; Apple silicon M1+Official site and X post
Operating systemsWindows 10/11, Linux and SteamOS with Proton, macOS 26+Official site
RTX 5090, 4K, 2XANR FG 1.01 ms / DLSS FG 1.26 msCompany site
RTX 5090, 4K, 4X meanANR FG 0.61 ms / DLSS FG 0.82 msCompany site
RTX 5090, 4K memoryANR FG 214 MB / DLSS FG 562 MBCompany site