Back to today’s briefing

Story perspectives

OpenAI's Models Outperform Open-Source in Token Efficiency

8/16/2025

38 6

Summary

Story summary

A Nous Research study shows open-source AI models require 1.5-4 times more tokens than closed models for the same tasks.
OpenAI's gpt-oss models exhibit better token efficiency, particularly in solving math problems.
The findings highlight the significance of token efficiency in choosing AI models for business applications.

1 of 1

Story summary

A Nous Research study shows open-source AI models require 1.5-4 times more tokens than closed models for the same tasks.
OpenAI's gpt-oss models exhibit better token efficiency, particularly in solving math problems.
The findings highlight the significance of token efficiency in choosing AI models for business applications.