I don't think they should have too much to worry about if they can bring out a transformative new model in the next year, but if GPT-5 is only slightly better than GPT-4 I would start to worry if I worked there - LLaMa, Gemini, Claude, Mistral are all <i>about</i> as good or a bit better than GPT. You can download LLaMa or Mistral for free, why pay OpenAI?<p>GPT3/4 were like pulling a rabbit out of a hat, it seemed at the time that OpenAI was guaranteed to be a huge player, Google or Apple-level. But now there are these competitors, where does the money come from? You can't burn tons of cash on running GPUs forever, and if they switch to a paid model for ChatGPT I think that 95% of people will switch to Claude or something that's free - because ChatGPT isn't that much better.<p>People were predicting that LLMs were going to be like a commodity and I think they've been proven right. A lot of companies can make them, and it's just GPU spend to keep them running. You need something to differentiate to make real profits.<p>Plus, I think the future of AI is going to be on-device to solve the issue of GPU spend. When manufacturers start making AI-specific hardware and every major OS has its own local LLM, where does that put OpenAI? How do they make money out of that?
Real question: Does this spell the end of the free ChatGPT? It would make sense. Make it free to get it in the public eye, get 'em hooked, then remove the free tier.
I am happy with perplexity, but I am considering the paid version of OpenAI.<p>Question: is anyone using the paid OpenAI Version for language learning? (oral Conversation)