OpenAI has officially launched GPT-4.1, a successor to the GPT-4o multimodal AI model released last year. The company introduced this new model during a live stream on Monday, highlighting several key improvements, including a larger context window and enhanced performance in coding and instruction-following tasks.
GPT-4.1 is designed to offer significant enhancements over its predecessor, GPT-4o. One of the most notable upgrades is the model’s ability to handle up to 1 million tokens of context, surpassing GPT-4o’s previous limit of 128,000 tokens. This expansion allows GPT-4.1 to process and analyze a more significant amount of input, including text, images, and videos, enabling more complex and accurate responses.
OpenAI claims that GPT-4.1 is not only more reliable than GPT-4o in handling long and short context lengths but is also better at ignoring irrelevant information. These improvements make GPT-4.1 a stronger option for a wider range of applications, particularly in tasks involving large-scale data analysis or intricate reasoning.
In addition to the flagship GPT-4.1 model, OpenAI has also introduced two smaller versions aimed at developers looking for more affordable options:
These smaller models are designed to bring GPT-4.1’s advancements to a broader audience, especially those who need lighter, faster models for specific tasks.
GPT-4.1 is priced 26% cheaper than its predecessor, GPT-4o, making it a more cost-effective option for developers and businesses. The introduction of these models also comes at a time when OpenAI is working to stay competitive in the rapidly evolving AI landscape, particularly with the release of DeepSeek’s ultra-efficient AI model, which has set a new benchmark for efficiency and cost-effectiveness.
As OpenAI moves forward with GPT-4.1, the company plans to phase out its GPT-4 model by April 30th, 2025. Similarly, the GPT-4.5 preview in the API will be deprecated on July 14th, 2025, as GPT-4.1 offers better performance at a lower cost and with reduced latency.
Despite the launch of GPT-4.1, OpenAI is already looking ahead to its next-generation models. GPT-5, which was initially expected to arrive by May 2025, has been delayed. CEO Sam Altman mentioned that the delay was due to the difficulty of integrating the technology smoothly. In the meantime, OpenAI is also preparing for the full release of its o3 reasoning model and the o4 mini reasoning model, with these expected to debut soon.
