The reason this makes sense is that orchestration is now possible—thanks to ChatGPT significantly lowering the price of its tokens. Team Lead: GPT-5.6 Sol Team Members: GPT-5.6 Terra, Luna, 5.3 Spark Workflow: 1) Develop an overall plan via chat on the ChatGPT web interface. 2) Share the plan with the team lead, then delegate tasks to team members—breaking them down into fine-grained assignments to avoid overloading any single thread. After individual development, consolidate and summarize findings in separate threads. 3) Conduct review and refinement afterward. This is my current development approach. I’m naturally someone who dislikes complex setups and believes simplicity is best—I’m also someone who avoids loops, money-related tasks, and anything tedious. This exact method suits me perfectly—and the results are excellent. Before version 5.5, I had to do a lot more work to achieve this; since 5.6, my workload has been drastically reduced. It’s fantastic. Plus, ChatGPT includes image generation capabilities, which allows us to produce significantly higher-quality images for development—a key differentiator from Entropic. @OpenAI, please consider resetting your pricing soon—and offer a 50% discount on annual Pro subscriptions. Say it nicely.