OpenAI launches GPT-4.1 model.

On April 15th, OpenAI introduced GPT-4.1, the next version of the GPT-4o multimodal AI model launched last year. During the livestream introducing the new model, OpenAI stated that GPT-4.1 has a larger context window and is superior to GPT-4o “in almost every aspect,” particularly with significant improvements in programming and instruction following.

OpenAI launches GPT-4.1, a powerful programming model. (Illustration: Getty Images)

GPT-4.1 is now available to developers, along with two smaller versions: GPT-4.1 Mini and GPT-4.1 Nano. The Mini version is more affordable for developers to test, while the Nano is even lighter, and Open AI claims it's the "smallest, fastest, and cheapest" ever.

All three models can handle up to 1 million contextual tokens – including text, images, or videos – far exceeding the 128,000 token limit of GPT-4o. “We trained GPT-4.1 to retrieve consistent scene information across the entire contextual length of 1 million tokens,” Open AI stated in a post introducing these new AI models. “We also trained it to be far more reliable than GPT-4o in identifying relevant text and ignoring distracting elements,” the company added.

GPT 4.1 is also 26% cheaper than GPT-40, a very significant figure after DeepSeek launched its ultra-low-cost AI model.

The launch comes as OpenAI is preparing to remove the GPT-4 model from ChatGPT at the end of April. The tech company claims that the upgrades to GPT-4 make it a “natural successor” to GPT-4, which was launched two years ago. With the launch of GPT-4.1, OpenAI also plans to stop supporting the preview version of GPT-4.5 on its developer API platform on July 14th. “GPT-4.1 delivers better or equivalent performance in many key features, with lower overhead and latency,” OpenAI stated.

Winning Friends

(Translated and compiled from The Verge)

Source: https://baodongnai.com.vn/xa-hoi/khoa-hoc-cong-nghe/202504/openai-ra-mat-mo-hinh-gpt-41-6be1424/