OpenAI’s new flagship model, GPT-6 Astra, is facing a wave of complaints on X just one week after release, according to Decrypt. Several developers said the model now feels weaker than it did at launch, with some describing a pattern of faster responses but lower-quality output. Anonymous developer synthwavedd wrote that Astra felt "noticeably dumber" and joked that it had undergone a "post-launch lobotomy." Developer Pranjal Paliwal, who had previously praised the model, said after reviewing code written by Astra that "we don’t have AGI, we have a performance regression." Pankaj Kumar suggested OpenAI may have lowered the model’s "juice value," meaning the amount of compute used before generating an answer. Others, including Salio and researcher Md Ismail Sojal, said tests using the same prompts produced worse results than on launch day. T3Chat founder Theo offered a different view, arguing that Astra may simply be less stable than Claude Fable and that users are now posting more bad outputs after the early honeymoon period ended. OpenAI has not yet issued a statement on Astra similar to the one it made earlier about GPT-5.6 Sol.
OpenAI’s flagship GPT-6 Astra is drawing criticism on X just one week after launch, with users posting that the model has become worse, according to Decrypt.
Anonymous developer synthwavedd wrote that 「Astra feels noticeably dumber today」 and joked that it was a case of 「post-launch lobotomy.」
Developer Pranjal Paliwal, who had previously praised the model, said after checking code written by Astra: 「We don’t have AGI. We have a performance regression.」
Pankaj Kumar described the symptoms as faster responses and lower-quality answers. He also said he suspected that OpenAI had reduced the model’s 「juice value,」 referring to the amount of compute used before generating a reply.
Salio and researcher Md Ismail Sojal compared Astra’s output on launch day with its current performance using the same prompts. Both said the newer results were worse.
T3Chat founder Theo offered a different explanation. In his view, Astra is simply less stable than Claude Fable, and users began sharing bad outputs in larger numbers once the honeymoon period wore off.
Decrypt also noted that OpenAI’s previous flagship, GPT-5.6 Sol, went through a similar cycle in July. At that time, OpenAI executive Tibo Sottiaux denied that the company had intentionally weakened the model, but said OpenAI had been experimenting with 「reasoning effort」 settings.
OpenAI has not released a similar statement about Astra so far.
This article was originally published by Bit.Fan. For more cryptocurrency news and market insights, visit www.bit.fan. Disclaimer:
The market information, project data, and third-party content displayed on this platform are for industry information sharing only and do not constitute any form of investment advice or return commitment.
Cryptocurrency trading carries high risks. Users should fully assess their risk tolerance and make independent decisions. All profits, losses, and legal responsibilities are borne by the users themselves.
