Claude3. ChatGPT. Grok. Gemini.
We are watching a very interesting race between different companies in the large language model (LLM) race. Most of your major technology companies are involved, with others lurking in the background.
These models are being created by the likes of Google, X.ai, and OpenAi. Then we have Amazon getting involved with a $4 billion investment in Anthropic. Microsoft is heavily tied to OpenAI.
And let us not forget Apple who hasn't fully released to the world what it is doing. That said, we cannot presume they are sitting idly by.
This leaves us with the last participant, and the one who made the latest move.
Earlier, Meta released Llama3, the latest update of its LLM. Based upon reviews, it pushed them to the front of the class.
This was accompanied by the release of a web interface, ai.meta.com.
Here are the specs from the blog post:
In going through some of the reviews, many had a great deal of success with the initial trials of the application. Personally, I messed around with the image generation.
Here are a couple Threads created using Llama3:
@taskmaster4450le/re-leothreads-exrfhrut
@taskmaster4450le/re-leothreads-swmcpbph
As we can see, Llama3 places a watermark on the image. This will help to keep people from trying to pass off these images as real.
One positive about what Meta is doing is this is open source. Many models are closed which means we need to have a counterbalance. Also, Meta appears to be opening everything. Some claim to be open source yet only open the weights.
Meta jumped in the lead for nOW.
Lurking in the background is Grok 2.0 from X.Ai and, of course, ChatGPT5, which many feel will blow the competition away when it is released.
Then we have Google which is one of the leaders in data and, likely, compute.
The key here is how quickly the ranking changes. Meta's lead will last up to the point that someone else releases an update. Whomever that is will jump to the front of the line.
It is how this is shaping up.
We could have someone else who is ahead come next week. There are so many players involved with high stakes.
Zuckerberg is not a well liked individual and for good reason. For many, he is pure evil. That said, in this instance, he is making some sense.
His view is that open source is the only way to go. While many claim that this technology is too powerful to have in the open since a nefarious actor could use it, the counterclaim is it might be too powerful to have in the hands of one entity.
If this is the path to AGI (or even ASI), which is hotly debated, is there any single entity, whether company or government, you trust, with this?
I think most would agree that is not a single entity. If this does indeed have the potential to end humanity, than we do not want Zuckerberg, Altman, Musk, or the nutcases at the NSA in full control of this without an opportunity to counterbalance it.
That is what open source does.
I will be going deeper into this in another article.
However, what we are watching is a race to the bottom. When it comes to LLMs, there is going to be no value. The money is going to reside, from this perspective, with the applications.
In other words, these LLMs are becoming commoditized.
Over time, as they get more advanced, few will need to be cutting edge. Most will suffice for the majority of use cases.
Another factor is going to be the the prices.
Right NOW, Meta is offering Llama3 for free. This could change. Nevertheless, if it is offering this much capability for zero cost to the user, why would most pay for other chatbots? Certainly there might be some specific use cases but the majority will find this to be enough.
None of this means it is a worthless pursuit. This is a high stakes game for these companies. The point is the monetization might not be as straight forward as it appears. Having 500 million people paying for chatbots is not the future.
Instead, we should focus upon what they are integrated into. Here is where the real money is going to come from.
For now, we can enjoy the race because it is advancing at a rapid pace.
