Here's the big one that was being passed around months ago, which nowadays usually gets dismissed out of hand because of when they did it (while ignoring the relative finding): https://metr.org/blog/2025-07-10-early-2025-ai-experienced-o...

Here's a slightly more recent one focused more on comprehension/learning than productivity: https://www.anthropic.com/research/AI-assistance-coding-skil...

Metr attempted to redo that first one to get trends over time, but couldn't recruit enough developers to get reliable results for it.