Michael Elabd(@MichaelElabd):As the agentic world moves towards more specialized workflows, efficiency becomes very important! A librarian who has been working at the same library for 30 years doesnt need to look up skills for how to answer a question or search through her tools to know that the library opens every day at 9 am! Same for agents! As agents specialize, they should become way more efficient at solving tasks. Thats why I am really excited about our research on intelligence density. The goal here being how do you get models to perform better on tasks with higher token efficiency!

2026.09.24 16:28

As the agentic world moves towards more specialized workflows, efficiency becomes very important! A librarian who has been working at the same library for 30 years doesnt need to look up skills for how to answer a question or search through her tools to know that the library opens every day at 9 am! Same for agents! As agents specialize, they should become way more efficient at solving tasks. Thats why I am really excited about our research on intelligence density. The goal here being how do you get models to perform better on tasks with higher token efficiency!

Trajectory@trajectorylabs

2026.09.24 16:19

There’s a good chance your open source model is costing more than the frontier. Cheap tokens ≠ cheap tasks. Here, we introduce Intelligence Density, and Density Aware Training, our post-training technique to achieve less wasted compute, better learning, all with no knobs to tune. Enabled by default in every Trajectory model.