Do OpenAI's models already follow the new Model Spec?

Not fully. OpenAI says the 29 October 2026 Model Spec describes the behavior it is working toward and that its models do not yet fully reflect it. OpenAI is testing new training methods meant to teach models the five core values and the reasons behind the Spec's rules, and plans to share results in later updates.

Can developers still change default behavior under the new Model Spec?

Yes. OpenAI's updated Model Spec keeps firm boundaries only where they protect people and society. Most of its guidance remains defaults that developers and users can override, ordered by an updated chain of command. Users and developers can explicitly ask the model to take a particular perspective, but the model should not take one just because it guesses that is what they want to hear.

Why did OpenAI remove style rules from the Model Spec?

OpenAI says style should follow what people prefer, across languages and cultures and for each person, using feedback and context as evidence. So the 29 October 2026 Model Spec retires most of its detailed style guidance. Stronger defaults stay in place, such as approaching questions from an objective point of view.

Can I reuse the text of OpenAI's Model Spec?

Yes. OpenAI dedicates the Model Spec to the public domain under the Creative Commons CC0 1.0 deed, and publishes its source in the openai/model_spec repository on GitHub. Each version has a dated page on model-spec.openai.com, and OpenAI runs a public feedback form for the 2026 update.

OpenAI · 2026-10-29 · major

OpenAI Model Spec update — five core values, and the model is now 'GPT'

OpenAI's Model Spec of 29 October 2026 reorganizes its rules for model behavior around five values, led by faithfulness to human intent. It adds guidance on context, human agency and side effects, and retires most style rules.

OpenAI artwork for the updated Model Spec announcement

OpenAI rewrote its public rulebook for model behavior around five values instead of a long list of rules.

Quick facts

Version	Model Spec 2026/10/29
Five values	Faithfulness to human intent, harm minimization, honesty and transparency, human agency, care
New guidance	Consider context carefully; support human agency; follow social norms
Retired	Most of the detailed style guidance
License	CC0 1.0 (public domain)

What is it?

The Model Spec is OpenAI's public description of how its models should behave, and the 29 October 2026 version reorganizes it around five values: faithfulness to human intent, harm minimization, honesty and transparency, human agency, and care. Alongside the values come new sections on reading context, supporting human agency and following social norms, plus wider guidance on side effects. The document now calls the model "GPT" instead of "the assistant".

How does it work?

Faithfulness to human intent sits at the base: OpenAI sets the boundaries, developers direct apps inside them, and users direct the work inside those, under an updated chain of command. Most rules are defaults that developers and users can override, with firm limits only where needed. OpenAI says the framing borrows from assistance-games research on alignment, and that it is testing new training methods to teach models the values behind the rules.

Why does it matter?

The Model Spec sets the defaults that ChatGPT and the OpenAI API work toward, so it shapes how the models treat instructions, memories and context. Dropping most style rules means tone should follow each user's preferences, while an objective point of view stays a firm default. OpenAI says its current models do not yet fully match the new Spec.

Who is it for?

developers building on OpenAI models and AI policy researchers

Frequently asked questions

Do OpenAI's models already follow the new Model Spec?: Not fully. OpenAI says the 29 October 2026 Model Spec describes the behavior it is working toward and that its models do not yet fully reflect it. OpenAI is testing new training methods meant to teach models the five core values and the reasons behind the Spec's rules, and plans to share results in later updates.
Can developers still change default behavior under the new Model Spec?: Yes. OpenAI's updated Model Spec keeps firm boundaries only where they protect people and society. Most of its guidance remains defaults that developers and users can override, ordered by an updated chain of command. Users and developers can explicitly ask the model to take a particular perspective, but the model should not take one just because it guesses that is what they want to hear.
Why did OpenAI remove style rules from the Model Spec?: OpenAI says style should follow what people prefer, across languages and cultures and for each person, using feedback and context as evidence. So the 29 October 2026 Model Spec retires most of its detailed style guidance. Stronger defaults stay in place, such as approaching questions from an objective point of view.
Can I reuse the text of OpenAI's Model Spec?: Yes. OpenAI dedicates the Model Spec to the public domain under the Creative Commons CC0 1.0 deed, and publishes its source in the openai/model_spec repository on GitHub. Each version has a dated page on model-spec.openai.com, and OpenAI runs a public feedback form for the 2026 update.

Try it

Read it at model-spec.openai.com/2026-10-29.html