OpenAI has started deploying two of its newer AI models—GPT-4.1 and GPT-4.1 mini—into ChatGPT, after the recent launch through the company's API products.
These newcomers come as the successors to the 4o model series, and possess performance enhancements with important improvements for software developers and IT companies.
With this launch, OpenAI is sunseting the GPT-4.0 mini from ChatGPT to pave the way for the advanced versions.
Designed to improve coding performance, instruction following, and long-context comprehension, the GPT-4.1 line introduces significant improvements over earlier generations. OpenAI highlighted the advances in a recent blog post: "These models beat GPT-4o and GPT-4o mini on everything, with significant improvements in coding and instruction following.". They also have bigger context windows—up to 1 million tokens of context—and can more effectively use that context with enhanced long-context understanding. They include an updated knowledge cut-off of June 2024.
Access and Subscription Details
In a notice posted on X (formerly Twitter), OpenAI announced that GPT-4.1 is now available to paid ChatGPT users via the model selector, including those under Plus, Pro, and Team plans. Free-tier users will not be able to access it, however, Enterprise and educational account holders can look forward to the update in coming weeks.
Improvements Over the GPT-4o Series
OpenAI has listed significant technical advancements in the GPT-4.1 series, indicating that it outperforms GPT-4o in many high-stakes benchmarks. The firm highlighted, "GPT-4.1 is dramatically superior to GPT-4o at a variety of coding tasks, including agentically completing coding tasks, frontend coding, making fewer gratuitous edits, adhering to diff formats consistently, maintaining consistent tool usage, and so on."
The lighter variant, GPT-4.1 mini, also produces impressive results. OpenAI says, "GPT-4.1 mini is a dramatic advance in small model capability, even outperforming GPT-4o on most benchmarks. It matches or surpasses GPT-4o on intelligence evals and cuts latency by almost half while decreasing cost by 83 per cent."
GPT-4.1 Nano: All About Speed and Efficiency
Though not yet part of ChatGPT, the GPT-4.1 nano is still accessible only through API. It is a lightweight and streamlined model intended for extremely efficient applications such as classification and text autocompletion. OpenAI terms it its "fastest and cheapest model available" and supports up to 1 million tokens of context. In terms of performance, it records 80.1% on the MMLU benchmark, 50.3% on GPQA, and 9.8% on Aider polyglot coding—surpassing even the GPT-4o mini in some areas.
OpenAI is targeting this lightweight version as the default option for developers who want speed and affordability without giving up much capability.
Read also| Elon Musk Warns: Sun's Expansion Will Eventually Destroy All Life on Earth
