Just one month after its launch, OpenAI is deploying its latest cutting-edge model, GPT-4.1, directly within ChatGPT. OpenAI had announced the GPT-4.1 model family (including GPT-4.1, GPT-4.1 mini, and GPT-4.1 nano) as a "significant step in the practical application of AI." Initially available via the API since mid-April, these models were designed with a particular focus on the real-world needs of developers, particularly in terms of coding, following instructions, and understanding long contexts.
What does GPT-4.1 bring to ChatGPT?
The improvements are notable, but they will interest a specific group of users. On the coding front, GPT-4.1 outperforms GPT-4o and GPT-4.5 (yes, it's hard to follow OpenAI's strategy for naming GPT models) and stands out for its ability to crawl code repositories, complete tasks, and produce working code. This new model is also more comfortable following instructions and outperforms GPT-4o in following technical and complex requests. OpenAI specifies that the GPT-4.1, mini, and nano models can now process up to 1 million tokens. This is a considerable advancement for GPT models, allowing them to manage huge code bases or numerous large documents. The models have been trained to process long context windows. They are therefore much more reliable than GPT-4o at identifying relevant text and ignoring distractors.
Vision is not left out for the GPT-4.1 family, which is improving in terms of image understanding. GPT-4.1 mini often outperforms GPT-4o on image benchmarks and the model shines when it comes to understanding long videos without subtitles.
Note that in addition to these performance improvements, GPT-4.1 models can be more cost-effective. GPT-4.1 is 26% cheaper than GPT-4o for median queries, while GPT-4.1 mini reduces latency by nearly half and cost by 83% compared to GPT-4o, while matching or exceeding its intelligence performance. The GPT-4.1 nano model is touted as the fastest and cheapest model available.
GPT 4.1: Who can benefit?
This major update is available to ChatGPT Plus, Pro, and Team users now, with Enterprise and Edu users coming soon. But that's not all: the miniature version, GPT-4.1 mini, replaces GPT-4o mini for all ChatGPT users, including those on the free version. While GPT-4o remains the default model for free users, GPT-4.1 mini will now serve as a fallback model when GPT-4o's usage limits are reached.
0 Comments