OpenAI's Tibo Sottiaux Says Early Testers Reviewed an Older Build of GPT 6.2 Sol and Astra
One of the funny reasons for the difference in vibes between the influencers and the crowd today on 6.2 Sol/Astra is that we gave a few folks access a while ago and then made a lot of changes in the last 3-4 days and I'm pretty sure no re-testing was done within those days and the videos were produced last week. Don't get me wrong, while the models are a hefty jump compared to their GPT predecessors, they are by no means perfect and there are a few things the teams are excited for the models to get better at in future iterations in terms of taste and niceness of use. Love all the feedback and we hear you!
Context
It explains why early influencer reviews and public reactions to GPT 6.2 differ, and signals that OpenAI is still tuning the models.
OpenAI launched GPT 6.2 Astra and GPT 6.2 Sol on Nov 17. Tibo Sottiaux works on OpenAI's Codex team. Some users had said their experience did not match early review videos.
Deep dive
Tibo Sottiaux of OpenAI posted an explanation for what he called the difference in vibes between influencers and the wider public on GPT 6.2 Sol and Astra. He wrote that OpenAI gave a few people access a while ago and then made a lot of changes in the last three to four days, and that he is fairly sure no retesting was done in those days, with the review videos produced last week.
He added that while the models are a hefty jump compared to their GPT predecessors, they are by no means perfect. According to the post, there are a few things the teams want the models to get better at in future iterations, in terms of taste and how pleasant they are to use. He thanked users for the feedback.
The post drew more than 1,400 replies, a sign of how much debate the launch has created. Independent results have been mixed: Artificial Analysis ranked GPT 6.2 Astra second on its Intelligence Index but found no gain on its coding agent index, and ARC Prize reported a 100% score on ARC AGI 3.
Sottiaux did not say what the late changes were or which early reviews were affected. The post is an explanation from an OpenAI employee, not a formal changelog.
What to expect next
Watch for OpenAI release notes that describe the late changes and any updates aimed at taste and usability. Reviewers who tested early builds may publish new comparisons on the shipped versions. Public benchmarks and user reports over the next week will show whether the gap between early impressions and wider reactions closes. OpenAI staff may also post more detail in reply threads, and any quick model update that targets taste or ease of use would show the company acting on the feedback Sottiaux described.
