The changelog highlighted that “[GPT-4o] consistently outperforms GPT-4 in various areas, including writing, coding, and STEM, during direct comparisons.” OpenAI also mentioned that recent enhancements have significantly boosted GPT-4o’s ability to follow instructions, solve problems, and facilitate conversations, making it a fitting successor to GPT-4.
GPT-4 was launched in March 2023 for both ChatGPT and Microsoft’s Copilot chatbot available online. Its various versions featured multimodal abilities, enabling them to process images and text—marking a first for OpenAI models widely used in the market.
According to OpenAI CEO Sam Altman, the substantial GPT-4 model incurred training costs exceeding $100 million. It was later succeeded by GPT-4 Turbo in November 2023, which is a more efficient and cost-effective alternative.
GPT-4 is currently involved in copyright disputes between OpenAI and various publishers, such as The New York Times. These publishers contend that OpenAI trained GPT-4 on their content without permission. In contrast, OpenAI argues that the fair use doctrine protects it from legal repercussions.
The retirement of GPT-4 is expected to align with the introduction of new models in ChatGPT. Reverse engineer Tibor Blaho has indicated that OpenAI is preparing a series of models named GPT-4.1—including GPT-4.1-mini, GPT-4.1-nano, and GPT-4.1—alongside the o3 “reasoning” model announced in December and another reasoning model designated as o4-mini.
