Join our daily and weekly newsletters for the latest updates and special content related to the industry’s leading AI coverage. Get more information
Openi is GPT-4.1 rollingIts new non -reasoning large language model (LLM), which balances high efficiency to Chat GPT users with low cost. The company is starting with Chat GPT Plus, Pro, and the team with its payments on the team, with the expected enterprise and access to the educational user in the coming weeks.
It also includes GPT-4.1 MINI, which transforms GPT-4O Mini as a default for all GPT users, including on the free terrain. The “mini” version provides a small scale parameter and thus, the less powerful version with similar protective standards.
The models are available through the “More Models” drop-down selection in the upper corner of the chat window inside both chat GPT, which provides users to select the argument models such as GPT-4.1, GPT-4.1 mini, and O3, O4-Mini, and O4-Mini-HIGH.
Initially, only third -party software and AI developers were planned to use by Open AI’s application programming interface (API), GPT -4.1 was added to Chat GPT after strong user feedback.
Open AI Post Training Research Lead Michelle Pokarus The shift confirmed on X was driven by demand, writing: “We were initially planning to keep this model API in the beginning but you all wanted it in Chat GPT 🙂 Happy coding!”
Openi Chief Product Officer Kevin Well posted on x Saying: “We have made it for developers, so it’s great after coding and recipes – try it!”
An enterprise -based model
GPT -4.1 was designed with ground -up for enterprise grade processes.
In April 2025, along with GPT 4.1 mini and nano, this model family preferred the developer’s requirements and production issues.
The GPT -4.1 scored 21.4 point improvements against the GPT4O on the SWE Bench -Certified Software Engineering Benchmark, and 10.5 points on the Scale Multi -Chhalage Benchmark at the Instruction Following task. This reduces the verb by 50 % compared to other models, a characteristic enterprise users are appreciated during the initial test.
Context, speed and access to model
GPT -4.1 Chat supports standard context for GPT: 8,000 tokens for free users, 32,000 tokens for Plus users, and 128,000 tokens for Pro users.
According to the developer Angel Boogado While posting on X, these limits first meet those used by Chat GPT models, though plans are underway to further increase the size of the context.
Although the API version of GPT -4.1 can take up to a million tokens, this expanded capacity is not yet available in Chat GPT, though future support has been indicated.
This extended context allows API users to feed the entire code base or large legal and financial documents in the model.
Open has acknowledged some of the performance degradation with the most large inputs, but enterprise test issues suggest solid performance up to millions of tokens.
Diagnosis and safety
Openi has also made a launch Safety evaluation center Website users to provide access to key performance measurements in the model.
GPT-4.1 shows concrete results in these diagnoses. In fact, in the tests of accuracy, he scored 0.40 on a simple QA benchmark and 0.63 on Prasanka, which performed many predecessors.
It scored 0.99 and 0.86 on more challenging signals in the Openi standard denial tests.
However, in a strong gel break test-Advocate terms, an educational standard-GPT-4.1 scored 0.23 runs, behind models like GPT-4 O-Mini and O3.
That said, he scored a strong score of 0.96 on the gests of humanity, which indicates the safety of a stronger real world under normal use.
In following the directive, the GPT -4.1 Openi’s default rating (System Over Developer, more than the user messages from the developer) follows the system with a score of 0.71 to resolve the user’s message disputes. It also performs well in protecting safe phrases and avoiding tuitioning scenes.
Make GPT -4.1 context against the predecessors
The release of GPT -4.1 has surfaced after scrutiny around GPT 4.5, which began in February 2025 as a research preview. This model emphasized on better undesirable education, a wealthy knowledge, and low deception-GPT4O fell from 61.8 % to 37.1 %. It also showed improvement in writing emotional newborn and long form, but many consumers found them rise.
Despite these benefits, GPT-4.5 criticized its high prices-API for poor performance in mathematics and coding benchmarks in mathematics and coding benchmarks against the PER 180 per million output token-and Openi Openi O-series models. Industry data noted that while GPT -4.5 was strong in general conversation and content production, it performed less in specific applications related to the developer.
On the contrary, GPT -4.1 is aimed at a fast, more concentrated alternative. Although it lacks the expansion of GPT-4.5 knowledge and widely emotional modeling, it is better to codify Better for practical coding aid and more reliable on user instructions.
On the API of Openai, GPT -4.1 is currently the price 00 2.00 2.00 million input token, 50 0.50 per million cathek input token, and .00 8.00 per million output token.
Available in L -G, GPT -4.1 mini -mini -input token of $ 0.40, 10 0.10 per million kitcard input token, and $ 1.60 per million output token.
Google’s Flash Light and Flash Model The base rate of GPT -4.1 is starting at $ 0.075- $ 0.10, starting at $ 0.075, starting at $ 0.40 per million output tokens per million, and 0.30 -$ 0.40 per million output token.
But although GPT -4.1 is high, it offers strong software engineering benchmarks and more precise instructions, which may be important for enterprise deployment scenarios, which requires more reliability. Ultimately, Openi GPT -4.1 provides a premium experience for precision and developmental performance, while Google’s Gemini model appeals to cost – -bearers, which requires flexible models and multi -modal capabilities.
What does this mean for enterprise decision makers
Introduction to GPT -4.1 brings specific benefits to enterprise teams managing LLM deployment, orchestations, and data operations.
- AI engineers monitor LLM deployment Can expect better speed and instructions to follow. The GPT-4.1 offers a more responsive and efficient tool set for full LLM Life Cycle Management teams from-Model Fine Toning. It is especially under pressure for lean teams to ship high -performing models without safety or compliance.
- AI Architeization Leeds Focusing on scaleable pipeline design will appreciate the strengthening of GPT 4.1 against most user -stimulating failures and its strong performance in messages Haier’s tests. This makes it easier to integrate into the archetypes system that prefers consistency, model verification, and operational reliability.
- Data engineers The GPT -4.1 will benefit from the lower fraud rate and the accuracy of the GPT -4.1 responsible for maintaining high data standards and connecting new tools. Its high prediction Output helps build reliable data work flu, even when the team’s resources are restricted.
- IT security professionals The responsibility of embedding security in the DOOPS pipelines can be worth the resistance of the GPT -4.1 joint gel brake and its controlled output practices. Although its academic resistance score leaves ROOM room for improvement, the model’s high performance against human -born achievements helps support secure integration in internal tools.
In these roles, the positioning of GPT -4.1 has been improved as a model description, compliance, and deployment performance of Optim, which makes it a great option for medium -sized businesses to balance performance with operational demands.
A new step forward
While GPT -4.5 represented the Skylling Single in the development of the model, GPT -4.1 utility centers. This is not the most expensive or most multi -model, but it provides meaningful benefits in areas that are important to businesses: accuracy, deployment performance and cost.
From this place to create the largest model at any cost and to make capable models more accessible and capable, reflect the industry’s wider trend. The GPT -4.1 meets the need, which offers a flexible, ready tool for teams, trying to embed the AI ​​in its business work.
Since the Openi continues to prepare its model offers, the GPT -4.1 represents a step forward in making the modern AI democratic for the enterprise environment. For decision makers, it provides a clear path to deployment without the ROI, to balance capacity, it without performance or safety sacrifice.
