OpenAI ChatGPT Removes Free User Chat Limits, Introduces Teen Safeguards

Post Views: 1

OpenAI enhances ChatGPT capabilities for free users while implementing enhanced protections for younger audiences

Model Enhancements for Premium Users

OpenAI has introduced updates to the GPT-5.6 Sol model, which underpins ChatGPT for Plus and Pro subscribers. The company has eliminated text conversation rate limits for free users, enabling uninterrupted interactions without waiting for cooldown periods. For premium accounts, GPT-5.6 Sol now integrates both rapid responses and extended analytical reasoning within a single model architecture, replacing the previous separation of Instant and Thinking modes. The updated model prioritizes focused, context-appropriate responses while minimizing extraneous formatting. A configurable slider allows users to adjust the depth of reasoning applied to specific queries.

Internal Testing and Accuracy Improvements

Internal testing on financial, medical, and legal prompts demonstrated significant improvements in accuracy. Results indicated a 62% reduction in factual errors with GPT-5.6 Luna and a 68% decrease with GPT-5.6 Sol compared to the prior GPT-5.5 Instant model.

Free User Access and Transition

Free users will transition to GPT-5.6 Luna as their default model this week. Starting next week, unlimited text chats will be available, though abuse prevention measures will remain active. File uploads, image processing, and other tools will retain existing restrictions.

Enhanced Safeguards for Younger Audiences

OpenAI has expanded safeguards for users under 18, training GPT-5.6 Sol to avoid romantic roleplay scenarios, age-inappropriate challenges, and content that could substitute human relationships. The model includes boundaries for sexual material, eating disorder discussions, body image risks, dangerous activities, and graphic violence. When detecting potential distress, the system prompts users to seek support from trusted individuals.

Security Evaluations and Resilience

Security evaluations confirmed GPT-5.6 Sol’s resilience against adversarial attacks. Testing against jailbreak attempts—prompts designed to bypass safety protocols—showed performance comparable to earlier models like GPT-5.5 Thinking. For prompt injection attacks, which attempt to override system instructions through tool outputs, GPT-5.6 Sol achieved a 1.000 score on standard evaluations, up from 0.998 for GPT-5.4 Thinking. On advanced search and function-calling tests, it scored 0.910.

Additional Updates and Ongoing Developments

Additional updates include enhanced resistance to malicious input techniques and ongoing refinements to improve youth-focused interactions. The company continues to develop evaluation frameworks to assess model performance for users under 18.