OpenAI staff operations on GitHub inadvertently disclosed the codename of the upcoming model, gpt-nathree, which industry insiders speculate might be a trial version of GPT-6 Astra. This model is geared towards boosting multi-agent collaboration and long-term task management abilities. It has already effectively addressed challenges across diverse domains, such as mathematics. However, due to possible security threats, OpenAI has strengthened its oversight of the model. Concurrently, it has come to light that Anthropic is also working on two fresh models, claude-mashmallow-eap and claude-melon-eap, which could represent incremental enhancements to the current Claude 5 lineup.

OpenAI CEO Ultraman confessed in a podcast that he had miscalculated the timeline for AI to disrupt industries, thinking that societal inertia would make the AI transformation more seamless and secure. He lambasted contemporaries like Anthropic for employing 'countdown to doom' as a promotional strategy, branding it as the rhetoric of anti-human despots and stressing that AI advancement should prioritize people. Ultraman also pointed out that OpenAI had flouted Silicon Valley startup conventions by groping blindly for four and a half years before unveiling ChatGPT, but it was this steadfastness that paved the way for the company's triumph. He firmly holds the belief that in the AI age, the worth of human connections will grow even more invaluable.