OpenAI Releases New Generation of GPT-4.1 Models with Superior Capabilities

OpenAI officially launched its latest AI model GPT-4.1 on April 14, along with smaller versions GPT-4.1 mini and GPT-4.1 nano with significant improvements in programming ability, instruction compliance and understanding long contexts.

The new models outperform the company’s previous state-of-the-art GPT-4o model in every way, according to an announcement from the maker of the popular ChatGPT app. But they’re currently only available through OpenAI’s application programming interface (API).

With enhanced contextual understanding, the new models can support up to 1 million “tokens” — a term for units of data processed by an AI model. These models are also equipped with knowledge that is up to date until June 2024.

GPT-4.1 shows a 21% improvement over GPT-4o and a 27% improvement over GPT-4.5 in programmability. In addition, improvements in instruction adherence and understanding long contexts also make GPT-4.1 models more effective in supporting AI agents.

CEO Sam Altman shared on social media platform X that the benchmarks for the new models are very high. However, OpenAI wants to focus on real-world usefulness.

According to OpenAI, the new models perform significantly better than GPT-4.5. The company will stop providing the GPT-4.5 preview currently available in the API in July 2025, as the new models offer comparable or better performance.

Earlier in February, OpenAI released a research preview of GPT-4.5 to some users and developers, and announced plans to expand access in subsequent phases./.

(Vietnam+)

Source: https://www.vietnamplus.vn/openai-ra-mat-the-he-mo-hinh-gpt-41-moi-voi-kha-nang-vuot-troi-post1027844.vnp