Whatever is going on with 4o, it’s also affecting GPT-4.5

Examining Recent Developments in AI Model Behavior: The Impact on GPT-4.5

In recent days, users have observed noticeable shifts in the performance and system prompts associated with OpenAI’s GPT models, particularly concerning GPT-4.5. While details are still emerging, these developments suggest that issues affecting GPT-4 are concurrently impacting GPT-4.5, prompting discussions within the AI community.

One noteworthy aspect is the alteration in the system prompt configuration for GPT-4.5. This change indicates that GPT-4.5 no longer operates under the same parameters initially intended, raising questions about the consistency and reliability of the model’s responses. Users have reported that, despite the branding, GPT-4.5’s output and behavior are not aligning as expected with the original GPT-4.5 model, leading to a sense of inconsistency.

For individuals and professionals relying heavily on these advanced language models—particularly those investing significant resources into access—such disruptions can be frustrating. Notably, some users are paying substantial monthly fees (up to $200) to access these services, making stability and predictability crucial for their workflows.

This situation underscores the importance of transparency from AI service providers regarding updates, changes, and potential issues affecting model performance. As the community awaits official fixes, users are encouraged to stay informed through official channels and share insights to collectively navigate these challenging periods.

We hope that OpenAI and its team address these concerns promptly, restoring the models to their expected functionality and ensuring continued trust and reliability for all users.