AdamHolterererinc/localllama7d agoIf you average the reasoning effort curves across a lot of different benchmarks, high looks like the most attractive reasoning effort for Fable 5.2#llm#reasoning effort#benchmarks#model performance10 comments