OpenAI’s New GPT 4.1 Models Excel at Coding

Open Announced that he left a new family today artificial intelligence Models are optimized for increasing their efforts to exit strict competition, which are increasingly stronger than companies, Google and Antropic. Models are available for developers via Openai application programming interface (API).

Openai is released from a three-dimensional model: GPT 4.1, GPT 4.1 Mini and GPT 4.1 nano. Head Product Officer Kevin Weil, a Lipestream, the most commonly used model of new models, better than GPT-4O, GPT-4.5, GPT-4.5, GPT-4.5, he said.

GPT-4.1 increased by 55 percent in a widespread widely used benchmark to measure the valor of coding models. The account is a few percent points from other Openai models. New models “Great during coding, excellent in complex guidelines, they are fantastic for construction agents,” he said.

The ability to write and edit the AI ​​models has significantly improved in recent months, allows you to improve the more automated ways and so-called prototipation AI agents. As opponents in the last few months Anthropical and Google Both provide models that are especially good in the writing code.

The arrival of GPT-4.1 was widespread in recent weeks. Openai probably tried the model on some famous leaderboards under the Pseudonony of Alpha Quasaria. Some users of the “Secret” model declare effective coding abilities. “Quasar, the other codex solved all the open problems I live in Genarated [sic] “An incomplete person via LLMS wrote in Reddit.

“Developers have a lot of care about coding and increased our model’s ability to write a functional code,” Michelle Pokrass working in open training during Livestream on Monday. “Trying to follow the code that will follow different formats and explore the repositories better, explore and compose management tests.”

All new models can analyze the 6-fold code at a time, which improves adjustments or ability to correct or fix errors. New models are better than the instructions provided by users to reduce the need to repeat commands in different ways to achieve any results. Openai demonstrated the demos of GPT-4.1, including a flashcard, including language learning.

GPT-4.1 is 40 percent faster than GPT.4O. The cost of interrogations has been reduced by 80 percent.

In the last few years, Openai showed parlayed feverish interest in Chatgpta remarkable chatbot opened in late 2022Access to a growing job in more advanced Chatbots and AI models. Last week in the TED interview, Altman, Openai was a 500 million-week active user and this use is “very fast growth”, he said.

Now Openai offers a smorgasbord of different models with different capabilities and different prices. The largest and most powerful model of the company’s GPT-4.5 has started in February, Openai, the product has been still called “research view” because the product is still experimental.