Will Anthropic, Google, xAI or Meta release a model that thinks before it responds like o1 from OpenAI by EOY 2024?