Openai officially released the GPT-4.1 and GPT-4.1 Mini models in ChatGPT. The company reported this in on X . The update affected both paid and free tariffs: subscribers Plus, Pro and Team gained access to GPT-4.1, and the light version of Mini is now available to all users. At the same time, the previous GPT-4.0 mini model is removed from the service, which is indicated in official notes for the release of GPT-4.1 .
New models are especially useful for programmers: GPT-4.1 should be more accurately followed by instructions, faster than GPT-4O, and more effective when writing and debugging code. Although the release took place back in April through the API, only now the models have become available directly in the ChatGPT interface.
However, the launch of the GPT-4.1 previously caused criticism in the research community. The reason is the lack of security report, which usually accompanies new models. In response, Openai explained that the GPT-4.1 is not the so-called “border” model (Frontier Model) and does not need the same deep assessment. according to the head of the security systems Openai Johannes Hideke, “GPT-4.1 does not introduce new modalities or methods of interacting with the model and does not exceed the level of O3 by level intelligence.
amid the release of Openai, promised more openness. The company launched a new section – Safety EvalUations Hub – and stated that it would more often publish the results of internal safety inspections of the AI Models.
On the same day, Bloomberg said that Openai is close to buying a Windsurf startup – one of the leaders in the AI tools for programming. The amount of the transaction may amount to $ 3 billion, . This will strengthen the Openai position against the backdrop of growing competition: in particular, Google has updated its Gemini chattorus by adding integration with Github.