Just as deepseek r1 + web search reasons based on web search results, the reasoning model in the assitant API should be able to move in and out of the vector store during the process of generating one answer to produce the correct answer.
Just as deepseek r1 + web search reasons based on web search results, the reasoning model in the assitant API should be able to move in and out of the vector store during the process of generating one answer to produce the correct answer.