A C++ Aimbot That Sells Better Weights on Patreon

What it does Sunone Aimbot 2 is a Windows application that watches your screen and automatically snaps your crosshair to enemy players in first- and third-person shooters. It runs YOLO object-detection models in .onnx format through either a CUDA/TensorRT pipeline for recent NVIDIA cards or a DirectML backend for broader GPU support. The project ships as a precompiled executable with an in-game overlay for tuning, and the README emphasizes that you do not need to compile the C++ source.

The interesting bit The project is less about novel computer vision and more about polished cheat infrastructure: it bundles WindMouse for “human-like” cursor curves and references KMBOX and MAKCU hardware input boxes, implying an effort to evade software-level input detection. Improved AI models are paywalled behind Patreon and Boosty subscriptions.

Key highlights

• Pre-built .exe for Windows 10/11 with two switchable inference backends: CUDA 13.1 + TensorRT, or universal DirectML.
• Supports modern RTX cards down to integrated Intel/AMD graphics, though TensorRT excludes GTX 10-series Pascal hardware.
• Overlay-driven configuration and on-the-fly model selection via the HOME key.
• References hardware mouse-emulation devices (KMBOX, MAKCU) and the WindMouse algorithm for humanized movement.
• Better-trained models are offered exclusively to Patreon/Boosty supporters.

Caveats

• Windows-only; no Linux or macOS support is mentioned.
• The CUDA/TensorRT build is locked to newer NVIDIA hardware (GTX 1660 and RTX 2000/3000/4000/5000) and requires specific proprietary driver libraries.
• README is largely a quickstart manual; architectural depth is left to the source code and sparse docs.

Verdict A reference point if you are researching real-time Windows game-hacking tooling or low-latency ONNX inference. If you are looking for a general-purpose vision framework or do not play Windows shooters, keep scrolling.