All jokes aside, I will vote for OpenAI being the company of the decade. The last company to bring this much innovation was Google in the yearly 2000, and even that does not compare to this. This could be compared to the Model T from Ford (first usable car).

Most of the time this works, however there's 3-4 occasions where it will stop 50% the way through for the page, add hundreds of spaces, and then add:

5. Opus often refuses to follow simple instructions 6. While GPT's responses may not be as concise, they are more consistent, probably more "production ready"

3. The same prompt gives very different results when used against the two models. You'd need to refine a prompt for each model 4. While Opus can show moments of brilliance, it can (and often) get things very wrong.

Here's my findings: 1. Opus is definitely a smarter model than GPT for these kind of tasks. It's responses are very in depth, concise, and overall intelligent. 2. Opus has huge potential to construct a knowledge graph from unstructured text. I'd say it was about 80% there

I've been playing around with Anthropic's new model Opus, for critical language-based tasks. These tasks included creating detailed and concise summaries on block text, and extracting entities and relations. I put every test head to head with GPT-4 Turbo.