DeepSeek Expose
I've had it up to here with the hype around this so-called revolutionary neural network "DeepSeek" that's "blowing up the charts." Initially, I thought it was just a case of sponsored media buzz, but the recent drop in NVIDIA's stock and the surge of news coverage in the CIS segment made me second-guess that assumption.
Let’s break it down and look at the facts:
If you directly ask the chat, "Are you using OpenAI's infrastructure?" (photos 2 and 3), it behaves oddly—there’s a long freeze followed by an error. We tested this on different devices and from multiple IPs.
Next, borrowing some thoughts from Igor’s channel (@seeallochnaya):
Is it really true that DeepSeek R1 is so much cheaper than OpenAI’s models? How does their business model even work?
Interesting Nuances
Here’s what most journalists don’t mention. Let’s head to DeepSeek’s Wiki page.
DeepSeek is a Chinese artificial intelligence lab that develops large language models with open-source roots. It is heavily funded by the Chinese hedge fund High-Flyer, founded and managed by Liang Wenfeng from Hangzhou, Zhejiang.
From this, two key conclusions arise:
They don’t necessarily need their models to be profitable.
They can generate income simply by releasing disruptive news that shakes the global markets.
If you check OpenRouter and compare DeepSeek’s pricing against competitors (since anyone can host models), you get an interesting picture (see photo). DeepSeek's model is hosted with a tiny context size, minimal throughput, and overall struggles to maintain quality (yellow status — performance degradation).
The Smoking Gun: Developer Documentation
When we dive into DeepSeek's developer documentation and look at photo 5, we find a blatant mention of OpenAI — and not just a mention but a direct dependency on the original OpenAI library.
Here’s a clear quote from their docs:
"Please install OpenAI SDK first: pip3 install openai."
This line makes it obvious: DeepSeek is entirely built on OpenAI's architecture and requires their SDK.
Igor's claim that they rushed this whole setup just to grab money appears true. They didn’t even bother setting up proper API keys.
ChatGPT's Opinion (Whose Infrastructure DeepSeek Uses and Takes Credit For):
Two Possible Outcomes:
1️⃣ DeepSeek is a hidden project by OpenAI: A secret testing initiative for new models (conspiracy-level theory).
2️⃣ DeepSeek was thrown together hastily: They use OpenAI’s infrastructure and blew their entire budget on marketing (the likely reality).
Final Thoughts
Thanks for your attention. Stay sharp and don’t let media manipulate you. If you share this post or its information, please credit me 😊