OpenAI releases new GPT‑5.6 prompting guide urging shorter, outcome‑first prompts

OpenAI has published a dedicated prompting guide for its latest model, GPT‑5.6 Sol. The guide advises developers to adopt outcome‑first prompting, define success criteria, and eliminate unnecessary instructions. It introduces a new text.verbosity API parameter and a section on Programmatic Tool Calling for bounded workflows. Internal coding‑agent tests reported that leaner system prompts improved evaluation scores by roughly 10‑15%, reduced token usage by 41‑66% and cut costs by 33‑67%. The guide warns that conflicting rules can destabilise the model and recommends avoiding absolute directives such as “always do this” or “never do that.” The full guide is publicly available on GitHub.