GPT-4.1 is a new line of Openai Shi Model created for programming

Openai announced A new series of artificial intelligence models is GPT-4.1, which includes three versions : full , GPT-4.1 mini and GPT-4.1 Nano . These models are not available in Chatgpt , but can be used through API companies. All of them have a context window volume of 1 million tokens, which allows you to work with text volumes up to 750 thousand words.

Openai says new models are particularly good for programming and executing instructions . According to the company, GPT-4.1 was improved thanks to developers' reviews. She is better writing a code for sites, makes less unnecessary changes, more precisely adheres to the desired format of answers and uses the tools you need more reliably.

GPT-4.1 competes with Google models ( Gemini 2.5 Pro ) and anthropic ( Claude 3.7 Sonnet ), which also have millions of context windows and show high results in benchmarks. GPT-4.1 showed up to 54.6%accuracy in the proven SWE-Bench Verified test, which is slightly lower than Gemini 2.5 Pro (63.8%) and Claude 3.7 Sonnet (62.3%). In Video-Mme, GPT-4.1 reached 72% of accuracy in the difficult category "long videos without subtitles".

Mini and Nano models are cheaper and faster, albeit less accurate. As for the value of:

Although GPT-4.1 demonstrates good results, it has its disadvantages. In particular, its accuracy is reduced when processing very long requests. In addition, the model often perceives the instructions too literally and requires clear and specific instructions. At the same time, OpenAi views GPT-4.1 as an important step in creating a SI agent capable of creating complete software products independently.