NVIDIA’s laptop RTX 5090 boasts sufficient VRAM to be able to crank every visual setting in the most demanding AAA titles, but when running AI workloads, it’s a whole new ballgame for the GPU’s 24GB framebuffer. Still, a series of tests shows that it’s able to hold its own against Apple’s M5 Max by comprehensively beating it in a series of LLM workloads, that is, until the benchmarks fire up a heavier AI model coupled with increased Context