I love obscure benchmarks, and I feel like I can trust their results a lot more - afterall, they (probably) weren't benchmaxxed. RuneBench[0] is another good ex

I love obscure benchmarks, and I feel like I can trust their results a lot more - afterall, they (probably) weren't benchmaxxed. RuneBench[0] is another good example (how well LLMs can play Runescape) [0] https://maxbittker.github.io/runebench/ https://maxbittker.github.io/runebench/