Navigating the Sea of LLMs

Jan 1, 2025

I saw a post from someone asking their followers to help them choose between ChatGPT and Gemini. Of course, they were likely doing it for engagement and the algorithm. But seeing other comments, I felt confident enough to leave mine. It wouldn’t hurt, as I’m actively building my personal brand.

This blog is inspired by that post and comment! I wanted to share my knowledge, tinkering, and insights about these hot topics as of writing.

The Players

The biggest players on the scene are ChatGPT by OpenAI, Gemini by Google, and Claude by Anthropic. Some notable others include Perplexity and Devin.

OpenAI

Founded in 2015 by several individuals, including Sam Altman, OpenAI’s mission is to ensure AGI benefits everyone - quite "sus," if you ask me. Initially a non-profit funded by the likes of Elon Musk, they transitioned to a capped-profit model. There’s so much money in AI that they actually held back from going fully for-profit! They dabbled in gaming and educational AI until 2020, when GPT-3 was released. That marked a pivotal moment in AI history.

Google (and DeepMind)

Most people know at least something about Google. They acquired DeepMind in 2014. DeepMind’s milestones include the groundbreaking AlphaGo (a Go-playing AI) and the introduction of the Transformer architecture, the backbone of modern LLMs. DeepMind also works on brain-related projects and general AI development across services like search, translation, and more.

Anthropic

Probably the least known among the three, Anthropic is a private research and development company founded more recently, in 2021. Their mission, much like OpenAI’s, is to build reliable and safe AI systems. What’s with these AI conglomerates and their lofty goals? Reportedly, Anthropic was founded in response to concerns about the pace and direction of AI development at OpenAI.

Advancements

CPUs, electricity, and GPUs are the same everywhere. Companies can only have so much of these toys. But software, on the other hand, is where the real differences lie.

The performance battles we see among these players are just short-lived campaigns for their products and profits. The trend is that they all surpass each other at some point. So there’s no reason to worry that one LLM or AI will permanently outshine the others!

Believe me, computing is expensive, especially at this level and scale. Here’s a simple guideline: if you’re concerned about AGI taking over, remember that it would require enormous computational power. Currently, the cost of such a system is upwards of $200/month. When that same power becomes available for under $10, then it’s time to raise the alarm. Until then, enjoy your tea and relax.

The Special Sauce

Each provider has something unique. If they were all the same, it would be boring. For example, Gemini integrates well with Google apps like Drive and YouTube. ChatGPT is built into Microsoft products like the Office suite. Meanwhile, Anthropic focuses on big and secure enterprise solutions.

Performance

As mentioned in the “Advancements” paragraph, one model might have a bigger context window, while another handles more tokens. Many factors contribute to these differences, but ultimately, there aren’t major disparities between them.

Sales & Marketing

The pricing and strategies are quite similar across the board. For instance, they all offer a certain number of “free” premium messages for a few hours. This uniformity has more to do with capitalism than technology. If one deviated too far from the others, it might face financial trouble or even bankruptcy.

Security & Morality

Ah, the notorious topic. It’s simple, really: big tech conglomerates are not entirely trustworthy. They have money, large legal teams, and close ties with governments. Exploiting citizens for data and other gains is logically inevitable. This is why it’s crucial to research the motives, scandals, and privacy policies of these companies. You won’t find a "good" one - just the least evil option.

It Depends, Of Course

If you’re struggling to decide, that’s okay. The point of this blog is to encourage you to research, explore, and understand the bigger picture surrounding these tools. Find out what features and trade-offs align with your needs, and choose accordingly.

What’s the Conclusion?

In the end, it doesn’t matter that much! What matters is that we maximize the benefits these tools offer. After all, they’re just tools - and will remain so for many more years to come!