OpenAI's Ted Sanders explains how its model version numbers signal release order and scale
Overview
OpenAI researcher Ted Sanders says the company's model version numbers indicate release order, relation to other releases, and the size of improvement.
He cites 6.1 as following 5.6 and not a large enough change to be named 7, according to Simon Willison's report. Sanders also says model size explains much of the variation in LLM performance, with larger models more reliable and more expensive per token.
Written by AI from the articles below · updated Oct 28, 8:04 PM ET
Check the sources:
Article timeline
The articles in this story. Times are ET.
- Simon WillisonBlogOpenAI's Ted Sanders explains how its model version numbers signal release order and scale
AITed Sanders, an OpenAI researcher, says model size explains much of LLM performance variation, with bigger models being more reliable and costlier per token. He says OpenAI's version numbers signal release order, relation to other releases, and magnitude of improvement, so 6.1 follows 5.6 and is not a major enough leap to be called 7.
Heat Trend
Not enough continuous observations to show a trend yet.