deep deep deep deep

At first glance, DeepSeek might seem like just another ChatGPT wannabe. But dig a little deeper, and things get more interesting.

Unlike most AI models that rely on supervised training, DeepSeek Zero uses reinforcement learning (RL) to build what it calls “robust reasoning abilities.” That’s tech talk for: it teaches itself through trial and error.

However, most users aren’t on DeepSeek Zero—they’re using DeepSeek-R1, which focuses more on readability and fixing language inconsistencies.

Now that the boring part is out of the way.

Let’s break it down: how does it stack up against ChatGPT?

1. Beating ChatGPT on the App Store

DeepSeek just snagged the #1 spot on Apple’s App Store for free app downloads. Impressive, right?

That being said, I can’t help but notice its logo bears an odd resemblance to the eerie Blue Whale Challenge design. Let’s hope that’s where the similarity ends.

2. Performance Issues

Despite its popularity, some users report that DeepSeek feels slow and buggy, with frustrating experiences like endless processing loops. Personally, I’ve seen it go “round and round” without delivering clear answers.

3. Pricing

This is where DeepSeek shines. It’s 27 times cheaper than OpenAI o1 while delivering comparable performance:

• OpenAI o1: $60.00 per 1M output tokens

• DeepSeek R1: $2.19 per 1M output tokens

For developers and businesses, that’s a huge difference.

4. Open Source & Self-Hosting

DeepSeek’s open-source design is a big selling point. And with GPU prices dropping, self-hosting AI could become the new normal. Imagine running powerful AI on your own hardware:

• No cloud dependencies

• No usage limits

• Complete privacy control

It’s like owning your favorite games instead of streaming them. You’re in control. Meanwhile, OpenAI keeps its models locked tight.

5. Geopolitical Concerns

There’s a catch, though. With TikTok facing bans and scrutiny, some worry that DeepSeek might face similar backlash. After all, it’s a Chinese product, and geopolitics can complicate things fast.

In between all this, one thing is clear: China is no longer just “catching up” in the AI race—it’s setting benchmarks. DeepSeek demonstrates that China’s AI ecosystem is ready to challenge global tech giants like OpenAI, Google, and Nvidia.

The result? More competition, faster innovation, and, yes, potential restrictions. Only time will tell.

We did a long discussion on DeepSeek so no quiz for today.

But here’s one thing I can’t leave without telling.

Chinese startup Lumos Robotics shared progress on LUS1 humanoid, which can now face hurdles, fall, and stand back up again.

Don’t forget—AI isn’t just growing online. We also have to consider how they might kick us down physically.

For now, given this demonstration, I feel safe.