The launch of OpenAI's long-awaited GPT-5 model last week marked an ignominious chapter for the company. While it boasted modest performance upgrades on paper, the actual experience of using it left many fans sorely disappointed.

Common criticisms were that its answers were too short and its writing was noticeably worse and devoid of personality. On top of that, many noted that despite its "PhD level" intelligence, it still made dumb errors like insisting there are three Bs in the word "blueberry."

It was such a letdown that fans demanded that OpenAI give them back access to the previous model GPT-4o, which the company boldly removed with the launch of GPT-5. And stunningly, OpenAI capitulated to fans' demands.

In sum, it did not feel like a "significant step along the path to AGI," as

See Full Page