OpenAI has announced the launch of GPT-4.1, an upgraded version of its GPT-4o multimodal AI model introduced last year. During a live stream event, the company highlighted that GPT-4.1 features a significantly larger context window and outperforms its predecessor in nearly all aspects, especially in coding tasks and following instructions effectively. This latest model is now accessible to developers, who also have the option to use two smaller variations: GPT-4.1 Mini, which remains cost-effective for developers, and GPT-4.1 Nano, which is described as the smallest, fastest, and most affordable model to date.
All three versions of GPT-4.1 can handle up to one million tokens of context, a significant increase from the 128,000-token limit of GPT-4o. OpenAI emphasizes that this new model has been trained to effectively focus on relevant information throughout the entire token range while filtering out distractions, making it considerably more reliable than its predecessor. Additionally, the cost of using GPT-4.1 is approximately 26 percent lower than that of GPT-4o, an important factor amidst the competition from DeepSeek’s ultra-efficient AI model.
The release of GPT-4.1 also coincides with OpenAI’s plans to phase out the two-year-old GPT-4 model from ChatGPT by April 30. The company indicated that recent enhancements to GPT-4o have made it an optimal successor for replacement. Furthermore, OpenAI will retire the GPT-4.5 preview in its API on July 14, asserting that GPT-4.1 provides better or comparable performance at a lower cost and latency.
This launch follows the announcement from CEO Sam Altman regarding the delayed release of GPT-5, which is now expected in a few months. OpenAI is also set to unveil its full o3 reasoning model and a mini o4 reasoning model shortly, with references already appearing in recent updates of ChatGPT.
