OpenAI has officially rolled out its latest artificial intelligence models, GPT-4.1 and GPT-4.1 mini, to ChatGPT users, expanding their availability beyond the company’s API for the first time. The new models, which had previously been restricted to developers through OpenAI’s programming interface, are now accessible directly within the ChatGPT platform.
In a recent announcement via its official X account, the AI research firm highlighted that the GPT-4.1 model is particularly well-suited for coding applications and complex instruction-following tasks. It is being introduced as a viable alternative to the o3 model, especially for users who prioritise technical performance and reliability.
These new models are being gradually made available to ChatGPT Plus, Pro, and Team subscribers through the “more models” menu. Enterprise and educational institution users will gain access in the coming weeks, OpenAI confirmed.
Meanwhile, free-tier users of ChatGPT will soon gain access to the GPT-4.1 mini model, which is set to replace the existing GPT-4o mini.
OpenAI describes GPT-4.1 as a major advancement in AI performance. In internal evaluations, the model scored 54.6 per cent on the SWE-bench Verified benchmark, highlighting its strong capabilities in software engineering tasks. It also demonstrated robust capabilities in instruction-following, with a score of 38.3 per cent on the MultiChallenge benchmark, and achieved 72 per cent on the Video-MME benchmark for multimodal long-context understanding.
A key feature of the GPT-4.1 series is its extended context window, which supports up to one million tokens, offering users the ability to process and reference vast amounts of information in a single interaction.
Notably, the GPT-4.1 mini model has also been shown to outperform GPT-4o in a range of intelligence evaluations. Despite its smaller size, it offers substantial advantages in terms of speed and cost, with OpenAI claiming it delivers results at half the latency and 83 per cent lower cost compared to GPT-4o.
