Today, a lab called Intology previewed Locus, an AI coding agent. It seemingly gets very impressive results, outperforming humans on METR's RE-Bench when no other AI system has even come close.

Could these results be too good to be true? (See discussion here.)

In six months (May 19, 2026), if Locus really seems to be about as big a deal as Intology's announcement would naively imply, this market resolves to YES. If not, or if things are still quite unclear, it resolves to NO.

It's possible that what Intology says about Locus's performance is literally true, but the implications are much more modest. For example, maybe Locus is extremely overfit to RE-Bench, such that it isn't much better than Claude Code at practical coding tasks. In this case, the market would resolve to NO.

Since this market resolves based on my subjective opinion, I won't trade on it.