Recently, Sam Altman of OpenAI and Dario Amodei of Anthropic have initiated gray testing for their latest models. Altman is overseeing the testing of GPT-6 Sol, while Amodei has bypassed version 5.1 and is directly rolling out Opus 5.2. According to user tests, Opus 5.2 has demonstrated notable enhancements in animation realism, front-end performance, and 3D capabilities when compared to its predecessor, Opus 5. These improvements bring it on par with GPT-6 Astra. At present, only users of Claude Code have access to the gray testing of Opus 5.2, with the web version predominantly featuring version 5.0. GPT-6 Sol can be distinguished by its knowledge cutoff date; if it can provide answers to queries regarding Opus 4.7, which was released on April 16, it indicates that the user is part of the gray testing group. Both companies harbor distinct strategies for the official launch of their new models. Amodei is leveraging the testing to generate buzz in anticipation of an Initial Public Offering (IPO), whereas Altman intends to introduce a more budget-friendly version and reinstate the $200 subscription service, a move that could be announced as early as this week. Furthermore, OpenAI has announced that the GPT-5.5 series will be discontinued on October 14. Additionally, registration for the OpenAI DevDay event, scheduled for late September, is now open, offering participants the chance to avail themselves of potential benefits.