OpenAI has introduced the GPT-4.1 and GPT-4.1 mini models to ChatGPT, enhancing the platform's capabilities for coding and instruction-following tasks. The GPT-4.1 model, initially released to developers via the OpenAI API in April, is now accessible to ChatGPT Plus, Pro, and Team subscribers through the 'more models' dropdown menu. This model excels in coding tasks and instruction following, offering faster performance compared to the previous GPT-4o model. For all ChatGPT users, including those on the free tier, OpenAI has replaced the GPT-4o mini with the GPT-4.1 mini. This new mini model serves as a fallback option for free users who have exhausted their daily usage limit of the default GPT-4o model. The GPT-4.1 mini is designed to provide improved performance at a lower cost, with an 83% cost reduction, making it an attractive option for everyday coding needs. The introduction of these models focuses on real-world developer needs such as coding, instruction-following, and understanding long contexts. The models support up to one million tokens, allowing for the analysis of large file uploads like entire GitHub repositories. In industry benchmarks, GPT-4.1 achieved a 54.6% score on SWE-bench Verified, a 21.4% improvement over GPT-4o, and a 38.3% score on MultiChallenge, a 10.5% increase. In coding tasks, human evaluators preferred GPT-4.1's outputs 80% of the time over GPT-4o.