The most exciting result to us is that our models behave qualitatively differently. In this example, our model immediately recalls relevant information in related cases from memory. It uses this information to reason through the answer in fewer tokens.

Our models study the workspace through thousands of rollouts before deployment. When solving new tasks, they produce better responses and get to the right answer faster. With study, we're able to outperform Opus 4.8 X-high while using 3.3x fewer tokens.

Today we're publishing our first research blog, Understanding a Law Firm through Study. We're sharing a glimpse of a future where agents are trained with native memory:

Thank you to @Nasdaq for supporting Engram on our launch day yesterday! Some have commented that this photo looks AI-generated. It's not. This really happened. Feel free to send this picture to your moms. We're certainly going to.

Today we announced our Initial Public Offering, a humble article on X. Thanks to the @NYSE for supporting us so early in our journey! And thanks to all of our lovely supporters here on X dot com for following along. More soon 🔜