OpenAI has initiated efforts to address performance degradation issues in its latest flagship model, GPT-6 Astra. Users reported significant declines in model capabilities shortly after its initial release, with many noting that the system frequently routes queries to lower-tier models despite active subscriptions.
The company has released a new version of ChatGPT and Codex, identified as 26.908.70816, to mitigate these concerns. While early feedback suggests that the update provides some relief, users indicate that the underlying performance issues have not been completely resolved.
Observers suggest that the degradation may stem from limited computing power or intentional system restrictions. International users have also reported frequent connectivity problems, further highlighting the strain on infrastructure following the launch of the new model.
To evaluate the model's performance, users are utilizing a dedicated testing platform that presents three random challenges. The analysis relies on identifying specific response patterns and stylistic quirks inherent to the model's training, which helps distinguish between different versions of the GPT architecture.
