"This is the best AI model yet"......Do we still care? Every company that releases a new model now claims that their AI model is the best. It's almost as if the rule is now "the latest is the greatest", and most people are getting disconnected from the hype. What AI companies need to realize is that people don't care about benchmarks and doctored graphs, they care about their problems getting solved. So, instead of telling me your AI model is the best thing since slice bread, why not tell me how well it can code and the type of coding tasks and projects it can handle. I am not talking leetcode-style stats but actual real-world developer work. Tell me how well your model can help me research a topic with hardcore facts to help my study, office or course work. Tell me how your model can help offload mundane tasks that I have to deal with on a day to day basis or how many languages it can now translate. This is why I believe specialized models will eventually win over the all-purpose language models being heavily pushed to consumers at the moment.