They are also pretty bad at navigating mazes (which can be somewhat similar in spirit to text adventures where you need to navigate through text): https://arxiv.org/abs/2507.20395
I stopped drinking coffee for four or five years. I drank a lot before. You do realise that it has a strong psychoactive effect, at least it did for me.
I still don't drink coffee, but I started experimenting with paraxanthine and I absolutely love it (paraxanthine is the primary metabolite of caffeine and is also a stimulant). I feel like it gives me most of the benefit of caffeine with very few downsides (no jitters, no crash, exits your system faster).
That's where the humanizers come in. These are solutions that take LLM generated text and make it sound human written to avoid detection.
The principle of training them is quite simple. Take an LLM and reward it for revising text so that it doesn't get detected. Reinforcement learning takes care of the rest for you.
It makes it look like the presentation is rushed or made last minute. Really bad to see this as the first plot in the whole presentation. Also, I would have loved to see comparisons with Opus 4.1.
Edit: Opus 4.1 scores 74.5% (https://www.anthropic.com/news/claude-opus-4-1). This makes it sound like Anthropic released the upgrade to still be the leader on this important benchmark.