The presenter tests Claude Opus 5.5 through Claude Code on browser puzzles and a browser-based physics simulation. The model completes several interactive challenges, but a parking puzzle takes more than ten minutes, which the presenter counts as a practical failure. Its first physics implementationCode generation uses AI or another automated system to create source code from instructions, examples, schemas, or higher-level specifications. scores poorly in a self-critique loop and freezes in the browser before further revision produces a usable simulationAgent completion verification checks evidence that an agent achieved the requested outcome rather than merely producing intermediate work..
Further demonstrations ask the model to perform a piano solo, build a mathematical explainer, reconstruct a property in Blender3D reconstruction estimates the shape, position and appearance of a three-dimensional scene or object from captured observations., create an Unreal Engine game and compose music through a desktop audio workstationComputer use lets an AI agent interact with software through interface controls such as screens, buttons, keyboards and pointers.. The presenter reports some impressive results, but the runs take substantial time; the property reconstruction is not fully faithful, and the final music track falls short of professional mastering. These are individual demonstrations described in the transcript, not controlled cross-model tests.
The review also identifies visual-recognition errors in the model's chat interface and surveys published benchmark and pricing claimsA benchmark is a defined test or collection of tasks used to compare a system's performance under specified conditions.. The presenter distinguishes those vendor or leaderboard results from his own trials. Together, the examples suggest substantial capability with meaningful limits in speed, fidelity and accuracy.
Watch on YouTube




