← All benchmarks

AidanBench (historical 62-prompt subset)

Chat & Writing · 2025-04-18

AidanBench's pinned Sonnet 3.7 non-thinking result sums valid answers across 62 prompts, omitting the prompt about spending more LLM output tokens for predictably better results. This incomplete raw total is retained as a diagnostic and is not comparable with the canonical 63-prompt total.

Top models (higher is better)