> if we'll see an explosion of purpose trained LLMs...

Domain specific models have been on the roadmap for most companies for years now for both competitive (why give up your moat to OpenAI or Anthropic) and financial (why finance OpenAI's margins) perspective.