An AI computing enthusiast has demonstrated Doom running on a custom virtual CPU reportedly designed by GPT-5.6 Sol, according to Tom's Hardware. The demo was shared by Angel (@Angaisb_) and shown inside Turing Complete, a computer-science education and puzzle game with a sandbox mode for building simulated computing systems. The outlet reports that the model named the processor design “Codex-R32.” In the video, the Doom viewport appears over a live, pulsing schematic of the CPU, exposing processor gates, registers, memory, arithmetic logic units and other blocks as the simulation runs. The technical claim is narrower than “AI built a real chip.” Tom's Hardware describes the system as a virtual CPU built from primitive logic components inside Turing Complete. The Doom build, according to an explanation Tom's Hardware attributes to the Codex, uses the C-based PureDOOM port compiled into native RV32IM machine code that runs directly on the custom Codex-R32 CPU. That makes the demo a hardware-and-software exercise inside a simulator: design a CPU from low-level components, compile a Doom port for its instruction target, then execute it visibly inside the sandbox. The same report says the display included live stats such as cycle count, simulation speed and memory/register values. Tom's Hardware frames the result as another entry in the long-running “can it run Doom” test for computing systems. Doom is a 1993 game, but it remains a compact and recognizable benchmark for whether an unconventional or constrained platform can execute meaningful interactive software. The caveat is performance. Tom's Hardware says the demo stretches the model and the sandbox, and that the result is not what anyone would call playable. In other words, the significance is not consumer usability; it is that an AI-assisted workflow reportedly produced a virtual CPU coherent enough to boot and run a Doom port at all. Because this cluster contains only one reputable report, the story should be treated as a reported demonstration rather than a broadly verified benchmark for GPT-5.6 Sol. The strongest supported takeaway is specific: Angel showed Doom running inside Turing Complete on a GPT-5.6 Sol-designed virtual CPU called Codex-R32, with the execution visualized at the schematic level. Who benefits: Educators, hobbyists and AI-tool builders benefit most if this kind of workflow becomes reproducible: it makes computer architecture more interactive and inspectable. Turing Complete also gains visibility as the environment used for the demonstration. Who's exposed: The exposed group is anyone tempted to read the demo as proof of real-world chip design capability. Based on the provided evidence, this is a simulator achievement, not a manufactured processor or a production benchmark.