What’s New in OpenAI’s GPT‑4.1? Bigger Context, Faster, Cheaper AI

Readout: OpenAI’s newest multimodal AI model GPT‑4.1 has been released.

OpenAI introduced GPT‑4.1 as the follow‑up to the GPT‑4o multimodal model, featuring a much larger context window and superior performance across almost all dimensions, especially in coding and instruction tracking.

GPT‑4.1 is now available to developers, along with two smaller, more affordable variants: GPT‑4.1 Mini, which is cost‑effective and easy to adopt, and GPT‑4.1 Nano, described as the smallest, fastest, and cheapest model to date.

All three models can process up to one million context tokens—far surpassing GPT‑4o’s 128 k token limit.

OpenAI stated that GPT‑4.1 was trained to reliably handle up to one million tokens and to more accurately focus on relevant text while ignoring distractions in both long and short contexts.

The new model is 26% cheaper than GPT‑4o, making it more cost‑effective compared with competitors such as DeepSeek.

This release coincides with OpenAI’s plan to phase out GPT‑4 by April 30 and to retire the GPT‑4.5 preview in July, positioning GPT‑4.1 as the natural successor with lower cost and latency.

GPT‑4o remains the default model in ChatGPT and recently received an update that added new image‑generation capabilities; the model’s popularity forced OpenAI to limit requests and pause free‑tier access to prevent GPU overload.

The announcement also hints at a broader shift in OpenAI’s roadmap, including a delayed GPT‑5 launch, upcoming full releases of the o3 and o4 mini inference models, and rumors of an internal social‑media platform prototype focused on ChatGPT image generation.