Navigating AI Updates in Business

When working with AI, it's really important to be familiar with multiple models and what they do best.

It's also important to note that AIs go through strategic seasons of growth and that not all updates are "updates."

For example, if you're using ChatGPT to run any key part of your business right now ... I'm sorry. I don't know what OpenAI is trying to accomplish with 5.1, but its memory is lower and its outputs are almost intentionally bad.

If you're good at what you do, you might find yourself correcting GPT 5.1 outputs more than keeping them -- unless you just want something you just said mirrored back in a more opaque voice, or you prompt like a programmer.

If you're a programmer, you may love 5.1 more than any other update.
If you're a creative, you might consider cancelling your subscription.
And if you're a business that has low-level employees using it to create outputs, December is going to be HUGE month to invest in the QA process.

Meanwhile, other LLMs like Claude, Perplexity, and Gemini are currently shining.

Claude remains the best writer,
Perplexity seems to handle information synthesis the best,
and Gemini is capturing the attention of AI artists as being a bit of an incomparable.

Then there's Grok, which seems to be updated almost daily and comes with the Grok Imagine feature that no one is really competing with.

But maybe the more interesting thing about Grok is that it seems to be trying to address the same problem ChatGPT is from a different position.

You see, both GPT and Grok are going through phases where their outputs aren't entirely useful in a straightforward way. But whereas GPT's are less useful because they are hyper-literal and increasingly dependent on single-prompt specificity, Grok's are less useful because they are inflated and hyperbolic.

They seem to be addressing the same problem from a different position: the tension between safety/reliability and creative usefulness.

GPT is zigging toward institutional trust and enterprise adoption, even at the cost of creative utility. Grok is zagging toward user engagement and cultural relevance, even at the cost of tripping into NFSW and unhinged outputs as a standard operating procedure.

They're both looking at the same problem, only one is turning the temperature up on the situation, while the other is turning the temperature down.

Neither approach is inherently wrong. They're just betting on different futures for AI.

GPT is betting that enterprises will pay premium prices for models that minimize liability and reputational risk, even if they're tedious to work with. Grok is betting that engaged users and strong brand loyalty will matter more than enterprise contracts, even if the outputs require more human oversight.

In the meantime, other LLMs make up ground by focusing on one thing and doing it best.

So, all that said: What LLM are you finding most useful right now?

Back to blog