Comment by aesthesia
8 hours ago
> Some of the claims categorized as "wrong" are also completely true, such as training hitting diminishing returns. New models are barely an improvement and most people I know stuck on Opus 4.6 over any newer one for example.
OK, but the first instance of a claim of diminishing returns was in February 2024, when GPT-4 was the best model available. Do you really think improvement since then has been minimal?
Yes, since 2024 the improvement has happened only in a few contexts, and the most famous¹ LLMs have also regressed in many contexts.
1 - Their numbers have also exploded, so I have no idea of any general rule.
I personally think improvement has been minimal since ChatGPT was released actually.
That's not defensible.