Anthropic is promising to do the same thing. They will release a blended model that will do the "usual" LLM and reasoning model in one. https://lnkd.in/gtd-AsNy
An interesting update from OpenAI. GPT-4.5 will be the last non-reasoning model, and o3 will be the last purely reasoning model. After that, they will merge it all into one GPT-5. So, test time compute won for good. The next big change I would love to see is integration of the long-term memory into the model itself. It is time for the RAG to die.
