Wikipedia has a list of language models that gives some information about release date, training size, parameter size, you can see that list here - https://en.wikipedia.org/wiki/List_of_large_language_models. I took that data and did some human curation on it giving a rough "capability" level to some of the models -

Download Sorted_AI_Model_Releases.csv

. I then asked GPT to graph those for me. I'm not too happy with the graph yet, the text is a bit hard to read, but I'll iterate on it a bit more later tonight.

The story here is that there has been a consistent flow of releases of increasingly powerful models, with new reasoning models coming online too. Tyler Cowen has a good take on what might be holding back the economic impact of these models - not too long, worth a read - https://marginalrevolution.com/marginalrevolution/2025/02/why-i-think-ai-take-off-is-relatively-slow.html?utm_source=rss&utm_medium=rss&utm_campaign=why-i-think-ai-take-off-is-relatively-slow.