In short
OpenAI printed a devoted prompting information for GPT-5.6 Sol that modifications earlier recommendation.
Inside coding-agent assessments confirmed lean system prompts improved eval scores by roughly 10–15%.
The information introduces a first-ever part on Programmatic Software Calling and highlights the textual content.verbosity API parameter—each absent from the GPT-5 playbook.
OpenAI printed a brand new prompting information for GPT-5.6 Sol, its newly launched flagship mannequin, and the primary message will really feel incorrect to anybody who spent the final 12 months writing multi-page system prompts: cease writing a lot. The core concept is outcome-first prompting. Outline what beauty like, set the stopping situations, and get out of the best way.
Detailed how-to directions, repeated model guidelines, examples that do not change habits—all of it’s now thought of noise.
OpenAI backs this with numbers: In inner coding agent assessments, leaner system prompts improved analysis scores by roughly 10–15% whereas slicing complete tokens by 41–66% and prices by 33–67%.
GPT-5 vs. GPT-5.6: What really modified
The GPT-5 prompting information, printed at launch in August 2025, was about including scaffolding. You bought XML persistence blocks telling the mannequin to maintain working till the issue was solved, detailed context-gathering templates that mapped precisely parallelize searches and when to escalate, and power preamble scripts that narrated each step out loud.
The philosophy was calibrating eagerness—constructing specific rails for when to go tougher or stand down.
GPT-5.6 largely would not want these rails. The brand new information tells you to trim: repeated guidelines, model directions that do not change habits, examples that do nothing, and course of steps the mannequin already handles reliably. So principally, that “ block with its parallel search batches and early-stop standards that used to assist is now scaffolding the mannequin has to parse round, not scaffolding that helps it.
What you really hold is easier: the user-visible consequence, success standards, stopping situations, and exhausting constraints. The information’s mannequin of a superb immediate begins with “Resolve the client’s difficulty finish to finish”—then specifies precisely what executed appears like, what actions to finish earlier than responding, and what to do when required proof is lacking. Not “be thorough.” Not “hold going.” Simply: right here is the vacation spot.
The danger calculus additionally shifted. The information warns that GPT-5.6 follows immediate contracts carefully, and that “conflicting guidelines can create extra instability than lacking element.”
An earlier mannequin would decide one instruction when it hit a battle. GPT-5.6 burns reasoning tokens attempting to reconcile each, which is slower, dearer, and infrequently incorrect. In case your system immediate has overlapping guidelines—and most manufacturing prompts do—that is the factor to repair first.
Additionally OpenAI closely advises in opposition to utilizing the previous trick of resorting to absolutes like “all the time do that” or “by no means do this” to steer the AI’s habits in a particular course.
Two concrete additions spherical out the distinction. The primary is the textual content.verbosity parameter: As a result of GPT-5.6 is already extra concise by default than GPT-5.5, previous “be transient” directions now over-correct and make responses too brief. Set a world default by way of the parameter, then override per activity within the immediate. The second is a piece on Programmatic Software Calling—for bounded workflows the place code handles filtering, batching, or aggregating giant intermediate outputs and returns a compact outcome, offloading that work from the mannequin’s judgment solely.
However does it work?
We used the information to optimize our immediate for TYPE OR DIE, the first-person typing survival horror recreation we construct to benchmark a mannequin’s coding talents. The outcome was extra polished: GPT-5.6 Sol tackled the auto-aim logic extra effectively than on earlier runs, the visuals had extra coherence, and the general really feel of the sport was cleaner.
It took extra time to construct. The mannequin did not bounce straight to code—it mapped all the drawback first, deliberate every system earlier than writing a line. That is the information working as meant. Outline the vacation spot; the mannequin chooses the route.
The brand new immediate is obtainable on our Github so you may test it out.
You possibly can play the unique GPT 5.6 recreation by clicking on this hyperlink.
The sport created beneath the newer immediate, is obtainable right here.
If you wish to push additional, or are too lazy to memorize all these new pointers, you may construct your personal customized GPT and feed it the total information as its information base. Configure it to research any immediate you throw at it, perceive the underlying logic, and rewrite it in GPT-5.6 model. You find yourself utilizing immediate engineering to engineer higher prompts.
Promptception. You are welcome.
Day by day Debrief Publication
Begin each day with the highest information tales proper now, plus authentic options, a podcast, movies and extra.
