A small company takes on Nvidia's DLSS with a free AI frame-generation beta that it says runs on older Nvidia, AMD and Intel GPUs

On the afternoon of Nov. 11, US Eastern time, a company called Apex Compute announced Apex Neural Rendering, or ANR, on X. It describes ANR as a cross-platform AI graphics package covering super resolution, ray reconstruction and frame generation, and claims it delivers "up to 2.4x faster performance and higher quality rendering than NVIDIA DLSS." Chinese tech site IT之家 (IT Home) and other outlets picked up the story the next day.

Some background. In big-budget PC games, how many images your graphics card can draw each second (the frame rate) decides how smooth the game feels. Nvidia's DLSS uses AI to take shortcuts. "Super resolution" renders the game at a lower resolution and uses AI to scale it up. "Frame generation" uses AI to guess an extra image between two real ones, so the frame rate looks roughly doubled. The catch: DLSS frame generation only works on Nvidia's newer RTX 40 and 50 series cards. AMD's FSR and similar tools work across brands, but many players feel their image quality trails DLSS.

ANR's pitch is that it works no matter whose graphics card you have. According to its GitHub page, the frame-generation beta supports Nvidia GTX 16 series and RTX 20 series or newer, AMD RX 5000 series or newer (including the Steam Deck handheld), and Intel Arc A and B series cards. It runs on Windows 10 and 11, and on Linux and SteamOS through Proton. The super-resolution and ray-reconstruction parts aren't out yet. The company says they're coming soon.

On the numbers, Apex Compute's own page says that on an RTX 5090 at 4K, ANR takes 1.01 milliseconds to generate each frame versus 1.26 ms for Nvidia's DLSS Frame Generation. In 4x mode it's 0.61 ms versus 0.82 ms, and video memory use is 214 MB versus 562 MB. It also published image-quality scores showing ANR ahead across the board. Keep in mind that the speed gap in the company's own table is about 1.25x, not the "up to 2.4x" in its X post, so the two claims don't line up. And every number and comparison image comes from the developer itself. Nobody has published independent tests yet.

Installing it is also more involved than ticking a box in a game menu. Per the GitHub instructions, you extract a package into each game's folder. It uses the open-source tool OptiScaler to stand in for Nvidia's frame-generation interface, and then you turn on the Nvidia frame-generation option in the game. In other words, it isn't officially supported by game developers. Results may vary a lot from game to game, and it could clash with things like anti-cheat systems. GitHub shows more than a dozen beta builds released in roughly the first day, so it's still changing fast.

Why it matters: in recent years AI frame generation has mostly been a perk of new graphics cards, and Nvidia has used it to nudge gamers toward upgrading. If a third-party tool can really match or beat it on older cards and rival brands, that ability would reach a huge number of existing PCs, and it would put competitive pressure on both Nvidia and AMD.

What it means for you: if you're running a graphics card that's a few years old, or a Steam Deck, you may get another free way to make games look smoother. For now it's a beta with a fiddly setup. Frame generation makes motion look smoother but doesn't make the game respond faster to your inputs, and it can even add a bit of lag, so be careful using it in competitive games. The safest move is to wait for independent reviews before trying it.

Original sources published: Apex Compute (X post), Nov 12, 2026, 04:44 Beijing time / Nov 11, 2026, 3:44 PM EST; Apex Compute GitHub first public beta, Nov 11, 2026, 17:50 Beijing time / 4:50 AM EST; IT Home (IT之家), Nov 12, 2026, 15:32 Beijing time / 2:32 AM EST