Ask HN: Is anyone giving out tokens for benchmarking LLMs?
I'm thinking of benchmarking different models performance playing Nethack and learning the game under different constraints.
1 thread
LMarena, now https://arena.ai. Not sure if they have API access to automate it though.