I get that Deep Seek is good, I use it all the time since ChatGPT 4o is stuck in 2023. But isn’t the hype a bit too much? The media acts like it’s the next big thing when really, it’s just built on the foundation OpenAI already made. Deep Seek R1, for example, is using ChatGPT’s base model. The only reason they could afford the $6M+ training cost is because OpenAI already spent billions creating the base.
Deep Seek is really just an optimized ChatGPT. It’s not leading the charge in AI, it’s just a polished version of what OpenAI has already done. We’re going to see more of these in the future—refined versions of the big models, not entirely new ones.
Again, I like Deep Seek, but I don’t think it’s groundbreaking. When ChatGPT 5 comes out, it’s going to dominate, and Deep Seek will just be catching up again.
Peyton said:
Saying Deep Seek is just a fine-tuned ChatGPT means you’re not looking at the full picture.
Deep Seek literally started as a fine-tuned ChatGPT. There was a point where it even identified itself as ChatGPT! The real magic is in how well they optimized it. It’s super efficient and cheap to run, and that alone makes it impressive.
I respect their work, but calling it a ‘new’ model is misleading. They took a strong foundation and made it better. But if OpenAI releases something better, Deep Seek will be playing catch-up again.
@Lin
DeepSeek-V3-Base is its own model. It uses a different approach with reinforcement learning, which is why it’s cheaper to run. It’s not just fine-tuning ChatGPT.
Deep Seek itself isn’t overhyped, but the way people talk about its effect on Nvidia, OpenAI, and Google is. People think Deep Seek is going to make Nvidia irrelevant by reducing training costs, but that’s not true. Nvidia’s business is built on the scaling law—more GPUs mean better models. Deep Seek doesn’t break that rule, it just lowers the cost a little.
OpenAI and Google already knew about this reinforcement learning approach. The only difference is Deep Seek made it public, so now everyone can use it. That’s a good thing, but the market freaked out and overreacted.
The real issue isn’t the tech—it’s the money. Investors thought OpenAI’s billions would keep competitors away. Now, Deep Seek shows that a strong model can be built for much less. That’s why stocks are dropping.
Deep Seek is open-source while OpenAI’s latest models are closed. That’s the real reason people are excited. You can run Deep Seek locally on your own machine.
Sam said:
Deep Seek is open-source while OpenAI’s latest models are closed. That’s the real reason people are excited. You can run Deep Seek locally on your own machine.
So is Deep Seek basically just ChatGPT but better?