OpenAI debuts its GPT-4.1 flagship AI model

OpenAI has introduced GPT-4.1, a successor to the GPT-4o multimodal AI exemplary launched by the institution past year. During a livestream connected Monday, OpenAI said GPT-4.1 has an adjacent larger context model and is amended than GPT-4o successful “just astir each dimension,” with large improvements to coding and acquisition following.

GPT-4.1 is present disposable to developers, on with 2 smaller exemplary versions. That includes GPT-4.1 Mini, which, like its predecessor, is much affordable for developers to tinker with, and GPT-4.1 Nano, an adjacent much lightweight exemplary that OpenAI says is its “smallest, fastest, and cheapest” 1 yet.

All 3 models tin process up to 1 cardinal tokens of discourse — the text, images, oregon videos included successful a prompt. That’s acold much than GPT-4o’s 128,000-token limit. “We trained GPT‑4.1 to reliably be to accusation crossed the afloat 1 cardinal discourse length,” OpenAI says in a station announcing the models. “We’ve besides trained it to beryllium acold much reliable than GPT‑4o astatine noticing applicable text, and ignoring distractors crossed agelong and abbreviated discourse lengths.”

GPT 4.1 is besides 26 percent cheaper than GPT-4o, a metric that has go much important pursuing the debut of DeepSeek’s ultra-efficient AI model.

The motorboat comes arsenic OpenAI plans to signifier retired its two-year-old GPT-4 model from ChatGPT connected April 30th, announcing successful a changelog that caller upgrades to GPT‑4o marque it a “natural successor” to regenerate it. OpenAI besides plans to deprectate the GPT-4.5 preview successful the API connected July 14th, arsenic “GPT‑4.1 offers improved oregon akin show connected galore cardinal capabilities astatine overmuch little outgo and latency. ”

GPT-4o, the existent default exemplary successful ChatGPT, was updated past period to bring new image-generation capabilities to the chatbot, which proved truthful fashionable that OpenAI had to bounds requests and enactment entree to escaped ChatGPT accounts connected clasp to prevent its GPUs from “melting.”

The GPT-4.1 uncover confirms our study from past week astir OpenAI preparing to motorboat caller models, and marks a pivot successful the company’s merchandise schedule. On April 4th, CEO Sam Altman announced connected X that the GPT-5 motorboat was being pushed backmost and should present get “in a fewer months,” aboriginal than the May deadline that was antecedently expected. Altman says that the hold is partially due to the fact that OpenAI “found it harder than we thought it was going to beryllium to smoothly integrate everything.”

OpenAI is besides acceptable to debut the afloat mentation of its o3 reasoning exemplary and an o4 mini reasoning exemplary immoderate time now, with references having already been spotted successful the latest ChatGPT web merchandise by AI technologist Tibor Blaho.