Musk says Digital Optimus plays Diablo by watching the screen

Elon Musk says Tesla's Digital Optimus can play about halfway through the Diablo campaign so far just by looking at the screen like a human. He posted it as a reply on X, with no video, no clip and no numbers attached.

Elon Musk

@elonmusk

X

Digital Optimus can play about halfway through the Diablo campaign so far just by looking at the screen like a human.

The same post says skill at Counter-Strike and other fast games is good too, that League is also in training, and that the goal is to generalize across all games. Musk describes it as interesting progress.

What was actually shown

Nothing. There is no footage, no success rate, no description of how many attempts it took to get halfway through the campaign, and no indication of whether the runs were picked from many tries. Nobody outside Tesla has seen it run. The entire story is one sentence from Musk and three short follow-up lines.

That matters because playing a game from pixels is exactly the kind of result that looks very different depending on the setup. An agent that reads the screen the way a human does, at human speed, with human inputs, is a different thing from one that gets privileged access to game state. The post does not say which this is. It says "just by looking at the screen like a human", and that is all the detail there is.

Why anyone cares

Digital Optimus is the part of the stack that does not need hardware. Games are a cheap, endless source of tasks where you can measure whether a policy did the thing or not, and where failure costs nothing. Getting one agent to play Diablo, Counter-Strike and League without being rebuilt for each is the generalization claim, and generalization is the whole argument for why a game-playing model has anything to do with a robot in a factory.

So treat this as a status update from Tesla, not a result. Until there is a video or a number, it is a claim in a tweet.