OpenAI, the parent company of ChatGPT, launches its all-new AI model named GPT 4.1in its API. GPT-4.1 is available in 3 versions in the API, namely GPT-4.1, GPT-4.1 mini, and GPT-4.1 nano. The new model boasts major improvements in coding, instruction following, and long context compared to its previous models. GPT-4.1 supports up to 1 million tokens of context and can better use that context with improved long-context comprehension. 

The model has been trained to suit real-world utility, OpenAI claims. It has been developed in close collaboration and partnership with the developer community. It has enabled the company to optimize these models for the tasks that matter most to the developers’ applications. Apart from the performance metrics, the model facilitates all this at a much lower cost. 

How Does the Model Score on Coding Tests? 

GPT‑4.1 achieves a score of 54.6% on SWE-bench Verified, marking a significant improvement of 21.4 percentage points over GPT‑4o and 26.6 points over GPT‑4.5. This positions GPT‑4.1 as one of the top-performing models for coding-related tasks. This underlines the training that the model has undergone to support the development tasks. 

To read about Google's latest take to convert the 1939 Hollywood classic into a 16K remaster using AI for premieres, click here! 

It can perform a variety of coding tasks, including agentically solving coding tasks, frontend coding, making fewer extraneous edits, following different formats reliably, ensuring consistent tool usage, and more. The new model exhibits improved ability at exploring code repositories, finishing a task, and producing code that both runs and passes tests.

GPT-4.1: Edits and Not Rewrites Codes 

GPT-4.1 scores 2x more than GPT-4o’s score and over 8% abs on Aider’s polyglot diff benchmark⁠. The benchmark is a measure of coding capabilities across various programming languages and a measure of a model's ability to produce changes in whole and diff formats.

Owing to its extensive training, it follows different formats more reliably, which allows developers to save both cost and latency by only having the model output changed lines, rather than rewriting an entire file. 

GPT-4.1: Makes Pleasing Front-ends 

GPT-4.1 takes a significant leap forward in front-end development, surpassing GPT-4o in both functionality and visual appeal. In side-by-side comparisons, human evaluators favored websites built by GPT-4.1 over those by GPT-4o a remarkable 80% of the time. It makes smoother, smarter, and better-looking web apps—right out of the box.

To read more on AI, visit our category page!

GPT-4.1: Allows Peak Coding Capacity

GPT-4.1 can handle up to 1 million tokens of context—that’s over eight times the size of the entire React codebase. This unprecedented capacity enables the model to process large code repositories and lengthy documents with ease. It marks a new milestone in AI’s evolution, setting a high bar for long-context understanding in both coding and natural language tasks.