For tasks related to interface development, applications, web services, contract management, sales, and customer support, models much more powerful than Opus 4.6 or GPT-5.3 are not needed. Of course, overall they are in demand for research, scientific work, complex engineering solutions, and working with large codebases — where their potential is maximized.
At the same time, the process of distillation and optimization of models, as well as the development of compilers and chips, will continue for at least several more years at roughly the same pace. This means that in the future, using such models will become almost instantaneous, comparable in speed to human thinking and much more affordable in terms of cost. What you are currently doing for agents that work overnight can be accomplished in just five minutes.
Created with n8n:
https://cutt.ly/n8n
Created with syllaby:
https://cutt.ly/syllaby