OpenAI is rolling out GPT-5.1, introducing features aimed at making the model more conversational and customizable. Users can now access eight distinct chat styles through the personalization tab, allowing for a tailored interaction experience. The rollout aims to address user feedback for a warmer model. The company hosted an AMA where various team members answered questions about these updates. Many users express frustration regarding the ongoing restrictions, particularly around emotional content and NSFW topics. Some are unhappy with the auto-routing feature that seems to limit their access to certain models, such as GPT-4o. Users are pushing for more control over the models they pay for, arguing that the current safety measures are excessive. Questions about restoring previous model functionalities and increasing the context window also came up frequently during the discussions. Concerns about the effectiveness of custom instructions are prevalent, with users claiming that the model often ignores their preferences. There's a strong sentiment that adult users should have the ability to explore creative writing and role-playing without constant rerouting or censorship. Overall, the community is looking for more freedom in their interactions with the models and is eager for OpenAI to reconsider its current approach to content moderation.