/research

Insights and blogs from the AI Group building Fin at Intercom

Think Fast: Reasoning at 3ms a Token

We step through the optimisation process required to make an Open Source reasoning model fast enough to use as a component of an interactive user application.

Read