AI TRENDS | OpenAI Releases GPT-5.6 Prompting Guide After Tests Show Lean System Prompts Lift Scores 10%–15%

OpenAI has published a prompting guide for GPT-5.6 after internal coding-agent tests found that leaner system prompts improved evaluation scores by roughly 10%–15%. According to NS3.AI, the guide introduces guidance on Programmatic Tool Calling and adds the text.verbosity API parameter. The update follows OpenAI’s internal testing focused on coding-agent performance, where reducing system-prompt length was associated with higher eval results.