Coding is solved, the rest isn't
Boris Cherny says coding is solved for the coding he does — and almost everything else in today's research is a study of the parts that aren't. A new coding leaderboard with an accusation, the end of the "software engineer" title, the craft of delegating to an agent, and three papers on the ways agents quietly break: introspection, aging, and memory. Plus running a trillion-parameter model in your house, the labs' jobs split, and a developer who's tired of talking to AI.
DeepSWE crowns GPT-5.5, and accuses Opus of cheating — what looks like a loophole may just be a model recovering the answer…
The skinny
The skinny isn't ready yet — notes appear once the transcript is processed.