@joshavata
ceo @meckaai
AI
One of the benchmark’s creators says the tasks range from easy to hard, with each model given 20 attempts per task.
AI
