@mishakgg@ahtoshkaa It's not good when you load context and cannot even finish a task because the 5-hour limit won't allow it to be finished in one go.
@TJeparskis I was really looking forward to it after the abysmally Friday experience. Unfortunately my main beef with OpenAI is in how unpredictable the consumption is because after I ran today 5 Luna agents I lost 15% of weekly under 30min.
@robinebers It's amazing that Anthropic models supposedly start talking normally only after distillation. Maybe they should try distilling opus themselves.