OpenAI and especially Anthropic have WEAKENED the models intelligence in novel research work.
Heat trend
Collecting trend data
The percentage is based on available heat signal, not comment count or independent people.
Claude model activity is surfacing — worth tracking for capability changes, ecosystem impact, and availability.
A user on reddit.com claims that OpenAI and Anthropic have "WEAKENED the models intelligence in novel research work." They state that last year's models allowed them to accomplish "10x the work" in novel research. The user finds current "smartest models" from Anthropic and OpenAI "useless" for achieving meaningful progress in research domains that might challenge these companies' dominance, questioning "What happened?"
I accomplished 10x the work with last years models doing novel research work. Anthropic and OpenAI smartest models are now the most useless when it comes to achieving any thing that can be attributed to meaningful progress in research domains that could potentially threaten their dominance.
Its really, really frustrating to be excited about the latest smartest models only to have them go FULL derp on topics they were actually decent at not even a year ago.
What happened?