OpenAI Launches GPT-4.1 on ChatGPT: Premium Users Gain Access to Its Most Powerful AI Yet

The update marks the arrival of OpenAI’s most advanced AI yet on ChatGPT, following an earlier rollout via API that began a month ago. In a post shared on X (formerly Twitter) on May 14, OpenAI confirmed that its flagship GPT-4.1 model is now live for ChatGPT Plus, Pro, and Team subscribers. Free-tier users will continue using GPT-4o, which OpenAI says will gradually inherit some improvements from the newer models.

What Makes GPT-4.1 Stand Out?

According to OpenAI, GPT-4.1 outperforms GPT-4o and its mini version across all key performance areas, particularly in software engineering tasks and instruction execution. In a blog post, OpenAI stated: “These models outperform GPT-4o and GPT-4o mini across the board, with major gains in coding and instruction following.”

One of the standout features is the significantly larger context window, which now supports up to 1 million tokens—allowing the AI to process and remember much more information than before. Additionally, GPT-4.1 features a more recent knowledge cutoff date of June 2024, compared to GPT-4o’s October 2023.

Benchmark results shared by OpenAI show a 21% absolute improvement over GPT-4o in software engineering tasks and a 10.5% boost in instruction following. These improvements make the model ideal for practical, real-world use cases such as writing assistance, customer service, and complex software development.

“While benchmarks provide valuable insights, we trained these models with a focus on real-world utility. Close collaboration and partnership with the developer community enabled us to optimise these models for the tasks that matter most to their applications,” OpenAI stated.

Lighter, Faster Versions Also Available

Alongside the full GPT-4.1 model, OpenAI has introduced GPT-4.1 mini and GPT-4.1 nano, designed to deliver high performance with reduced costs and latency.

• GPT-4.1 mini cuts latency by nearly 50% while being 83% more cost-effective than GPT-4o, making it a practical choice for applications needing speed and efficiency.

• GPT-4.1 nano is OpenAI’s most lightweight and affordable option, ideal for simpler tasks such as text classification or autocomplete functions.

“These models push performance forward at every point on the latency curve,” OpenAI noted in its blog post.

Who Can Access GPT-4.1?

Currently, the new GPT-4.1 series is only available to ChatGPT Plus, Pro, and Team users. Free users won’t have access for now but will continue using GPT-4o, which will gradually benefit from some of the newer model’s improvements.

For developers and businesses, GPT-4.1 is also available via API. OpenAI has introduced significantly reduced pricing to make the new models more accessible. GPT-4.1’s input costs now start at $2 per million tokens, while GPT-4.1 nano is priced from just $0.10 per million tokens. Additionally, prompt caching discounts have been increased to 75%, helping reduce costs for repeated queries.

Phasing Out Older Models

As OpenAI rolls out GPT-4.1, it is also retiring some earlier versions. The GPT-4.5 Preview, launched primarily for research purposes, was removed from the API on April 14, 2025. The original GPT-4, which had powered ChatGPT Plus since March 2023, has also been officially discontinued.

Although GPT-4.1 does not replace GPT-4o within the ChatGPT interface, many of its capabilities will be integrated into the GPT-4o experience over time. For those seeking top-tier performance, subscribing to a paid ChatGPT plan or accessing GPT-4.1 via API is now the go-to path.

As OpenAI continues to refine and scale its models, GPT-4.1 marks a significant step forward in making generative AI more powerful, efficient, and practical for real-world applications.