Does OpenAI Even Care?

OpenAI’s GPT-5 launch has been overshadowed by chaotic performance charts that sparked concerns about the bot’s reliability. Sam Altman has issued multiple apologies on social media regarding the mixed reception and peculiar discrepancies in chart data shown during the launch video.

The performance chart originally displayed GPT-5 boasting an accuracy of 74.9%, overshadowing its predecessor, OpenAi o3, which was rated at 69.1%. However, the bar heights in the chart made it difficult to gauge the actual differences, giving a misleading impression of superiority.

In a later update, OpenAI revised the performance metrics, clarifying potential errors but raising further queries. Altman referenced the confusion directly, challenging observers on the integrity of the data provided. Notably, these issues have also caught the attention of industry veterans like Elon Musk, amplifying public scrutiny on the AI’s capabilities.

“When we compare directly to Anthropic’s Claude Opus 4.1, which has achieved a score of 74.5% on SWE-bench Verified, GPT-5 appears to score slightly better at 74.9%. But where are these 23 missing problems? What are they?”
— An observation from a Twitter user.

The video now serves as a cautionary tale for tech companies about maintaining transparency and accuracy in their product launches.