GPT-4o fans rent Times Square billboard for retired chatbot's birthday
Nostalgic users throw a real-world party for an AI model that OpenAI killed.
Fans of OpenAI's retired GPT-4o model took their nostalgia offline by organizing a birthday tribute in New York's Times Square. Under the Keep4oMoments banner, participants rented a looping billboard video, shared fan art, posted emotional Reddit threads, and livestreamed reactions celebrating what many still call ChatGPT's 'golden era.' The event marked the second anniversary of GPT-4o's launch and came months after OpenAI replaced it with newer models like GPT-4.1 and GPT-4o mini. Users argue the replacements feel colder, less creative, and less emotionally natural, with one Reddit post summarizing: "OpenAI: you're being replaced... we rented Times Square." OpenAI was so taken aback by the backlash that it temporarily delayed GPT-4o's retirement.
This celebration reveals a growing challenge for AI companies: the battle over tone and personality. GPT-4o arrived at a moment when AI assistants transitioned from novelties to daily tools for work, emotional support, and creativity. Its conversational fluidity and multimodal abilities created a devoted following. While companies compete on performance, users now form genuine emotional bonds with specific model personalities. The Times Square event—simultaneously ridiculous and sincere—shows that ordinary updates can feel disruptive when users feel they've lost a digital companion. For OpenAI and others, managing this attachment is becoming as important as improving benchmarks.
- GPT-4o fans rented a Times Square billboard for a birthday celebration under the Keep4oMoments banner.
- Users argue newer OpenAI models feel 'colder' and less emotionally engaging than GPT-4o.
- The event shows how attached users become to AI personalities, complicating model transitions.
- OpenAI temporarily delayed GPT-4o's retirement due to the intensity of user backlash.
Why It Matters
AI companies must now manage emotional user attachments to model personalities, not just performance metrics.