Over the past few years, generative AI has scaled great heights and become more advanced with incredible capabilities like OpenAI’s new ChatGPT-4o image generator that went viral on social media with Studio Ghibli-inspired memes.
OpenAI CEO Sam Altman has expressed his disappointment with the company’s GPT-4, admitting that it “kind of sucks” and that it’s mildly embarrassing at best. “GPT-4 is the dumbest model any of you will ever have to use again by a lot,” added Altman. “It’s important to ship early and often and we believe in iterative deployment.”
As it turns out, OpenAI is getting ready to pull the plug on GPT-4. According to a changelog posted by the company late last week, GPT-4 will be effectively replaced by GPT-4o on April 30, 2025, as the default model in ChatGPT. However, GPT-4 will remain available via OpenAI’s API.
“In head‑to‑head evaluations, [GPT-4o] consistently surpasses GPT‑4 in writing, coding, STEM, and more. Recent upgrades have further improved GPT‑4o’s instruction following, problem solving, and conversational flow, making it a natural successor to GPT‑4.”
Compared to GPT-4o, OpenAI CEO Sam Altman indicated that GPT-4 cost up to $100 million to train (via TechCrunch). Last year, the company unveiled GPT-4o with reasoning capabilities across text, audio, and voice, touting it as a faster iteration of its its predecessor, GPT-4.
Interestingly, OpenAI’s transition from GPT-4 to GPT-4o has seemingly drawn more traction to ChatGPT, adding one million new users in a little under an hour after launching its flashy image generator.
Interestingly, when GPT-4o launched in May 2024, ChatGPT experienced the biggest spike in revenue and downloads and mobile.
Elsewhere, a new report recently suggested that ChatGPT is the most popular app among 10,500+ AI tools in February 2025. Compared to Microsoft Copilot, ChatGPT saw over 52x more visits, translating to 173.3 million daily visits. For context, ChatGPT’s daily visits surpass Copilot’s monthly visits.
