Claude vs GPT-4 vs Gemini for Autonomous Agent Tasks: My Production Benchmark

I spent three weeks and about $340 benchmarking three LLMs on the actual tasks my autonomous agents...