🤖 OpenAI brings back GPT4o 🤖
AI companion revolution emerges as users demand emotional connection over raw performance

OpenAI is bringing back GPT-4o less than 24 hours after replacing it with GPT-5, following an intense backlash that revealed the depth of human-AI emotional bonds.

Users flooded Reddit, X and social media with complaints about GPT-5's "flat" tone, with many literally shedding tears over losing their "AI friend." The viral hashtag #Keep4o dominated X as users shared stories of psychological support and emotional attachment to the model's personality. Many called the new model "lobotomized" and threatened subscription cancellations.

Sam Altman promised to restore GPT-4o access for Plus subscribers, though the default model will still be GPT-5 admitting he "underestimated" how much personality traits mattered. This backlash signals we're entering the AI companion revolution - and we're already living it without fully realizing the psychological impact.

💻 GPT-5 delivers breakthrough vibe coding capabilities 💻
Users can now build fully working apps one shot prompt

GPT-5's standout feature proves to be its revolutionary vibe coding abilities, allowing users to create fully functional applications using only natural language descriptions.

During live demos, OpenAI showed GPT-5 building a French language learning app with Snake-like gameplay in minutes, complete with progress tracking and quizzes. Mashable's testing confirmed dramatic improvements over GPT-4o, with GPT-5 successfully creating an interactive Titanic sinking visualization with timeline controls and clickable information panels.

The model can now work autonomously for extended periods, making "agentic coding" a reality for non-programmers.

🎨 Genie 3 turns famous paintings into explorable worlds 🎨
AI world model goes viral on social media as users step into masterpieces

Users on social media discovered a new viral use case: transforming famous paintings into interactive 3D worlds that users can explore in real-time. The AI world model, originally designed for general-purpose environment generation, allows users to literally step into Van Gogh's "Starry Night" or walk through Monet's gardens.

Videos of users exploring these painted worlds have exploded across Instagram, TikTok, and Youtube with creators showcasing surreal experiences of navigating through brushstrokes and artistic perspectives. Genie 3 generates these interactive environments at 24 frames per second in 720p resolution, creating an unprecedented bridge between classical art and immersive technology that's capturing social media's imagination.

INTERESTING TO KNOW

⚡Tesla abandons Dojo project for AI5/AI6 chips ⚡

Tesla has officially ended its Dojo supercomputer project, citing internal inefficiencies and significant leadership turnover including the departure of architect Peter Bannon. Elon Musk confirmed the strategic pivot, stating Tesla will focus resources on AI5 and AI6 chip development through Samsung's Texas fabrication facility, with AI5 chips expected in late 2026. The decision follows a talent exodus that saw around 20 Dojo team members leave to start DensityAI, a data center services startup.

🤖 Humanoid robots still lack AI breakthrough, Unitree CEO warns 🤖

Artificial intelligence technology sufficient for mainstream humanoid robot deployment remains a critical challenge, according to Wang Xingxing, CEO of prominent Chinese robotics developer Unitree.

Speaking to Bloomberg, Wang estimated the necessary AI expertise could be achieved within one to three years, comparing the current environment to ChatGPT's breakthrough moment in 2022. The assessment comes as Unitree continues making headlines with hardware innovations, highlighting the gap between mechanical capabilities and AI intelligence needed for real-world applications.