This article may contain affiliate links. Learn more on our Partner Program page.
OpenAI has released new language models. They are called GPT-4.1, GPT-4.1 mini, and GPT-4.1 nano. All three are available via the OpenAI API but are not yet integrated into ChatGPT. The models focus primarily on improvements in programming tasks, longer contexts, and better command tracking.
Anyone who regularly works with AI—whether in coding, analyzing large amounts of text, or in the field of automation—knows the weaknesses of many models: too slow, too expensive, or lacking sufficient contextual insight. With GPT-4.1, OpenAI aims to improve precisely these aspects. The new versions are designed to be more powerful, faster, and less expensive than their predecessors.
Three new models for different applications
OpenAI introduces three new models with the GPT 4.1 release:
- GPT-4.1: The main model. It performs best, especially on coding tasks and long contexts.
- GPT-4.1 mini: Slightly slimmed down, but faster and cheaper. According to OpenAI, it outperforms GPT-4o with equal or better performance.
- GPT-4.1 nano: The smallest and most affordable model. Designed for simple tasks such as classification or autocompletion.
The models are currently only available via the OpenAI API. They are not yet integrated into ChatGPT itself. However, OpenAI says that some of the improvements have already been incorporated into GPT-4o, with more to follow.
Strong results in coding and context processing
A major focus is on programming tasks. According to OpenAI, GPT-4.1 performs 21.4 percent better than GPT-4o and 26.6 percent better than GPT-4.5 in the SWE-Bench Verified Benchmark. Particularly interesting is that GPT-4.1 mini achieves similar or even better results than GPT-4o – at significantly lower costs and lower latency. Another aspect is the context scope. GPT-4.1 can hold up to one million tokens in context. This means that the model can understand even very long texts or large codebases coherently without losing track. This is a real advantage for many practical applications (via press release).
Prices at a glance
The three models differ significantly in price:
- GPT-4.1: $2 per million input tokens and $8 per million output tokens
- GPT-4.1 mini: $0.40 per million input tokens and $1.60 per million output tokens
- GPT-4.1 nano: $0.10 per million input tokens and $0.40 per million output tokens
OpenAI thus offers flexible options – depending on whether performance, speed, or cost are paramount. According to OpenAI, customized models are also available, but they cost more.
Current models and discontinuation of GPT-4.5
All new models are based on the June 2024 version. The model doesn't recognize events after this date. With the release of GPT-4.1, GPT-4.5 will be replaced in the API. The reason: GPT-4.1 offers almost the same features – at a lower cost and with better performance.
More performance, less cost: This is GPT-4.1
With GPT-4.1, OpenAI is launching three new models that excel in coding, long contexts, and efficient command processing. Users can choose between high performance (GPT-4.1), a good compromise (GPT-4.1 mini), and a fast, affordable solution (GPT-4.1 nano). GPT-4.5 is being removed from the API because GPT-4.1 offers more for less. If you work with AI in development or data analysis, these models are worth a closer look. (Image: OpenAI)
- Set up ChatGPT as a search engine in Safari – Here's how
- New ChatGPT update: Code editing in Xcode & more
- GPT-4.5 for ChatGPT introduced – This is what the new model contains
