OpenAI announced GPT-4.1, GPT-4.1 mini and GPT-4.1 nano for the API on April 14, 2025, and said the series targets coding, instruction following and long-context comprehension.
OpenAI said the GPT-4.1 models support up to 1 million tokens of context and have a refreshed June 2024 knowledge cutoff.
OpenAI said GPT-4.1 is available only through the API, while ChatGPT receives many of the same instruction-following, coding and intelligence improvements through GPT-4o updates.
The OpenAI model documentation lists gpt-4.1 as the model ID, describes it as the smartest non-reasoning model, and says it supports text and image input with text output.
The model documentation lists the default snapshot as gpt-4.1-2025-04-14, a 1,047,576-token context window, 32,768 maximum output tokens and a June 1, 2024 knowledge cutoff.
The same model page lists Chat Completions and Responses as supported endpoints, and lists streaming, structured outputs, predicted outputs, function calling, file search, image input, web search, fine tuning and prompt caching as supported features.
OpenAI's prompt-engineering guide tells API developers to pin applications to specific model snapshots, including gpt-4.1-2025-04-14, when they need consistent behavior.
The prompt-engineering guide also says developers can use the instructions API parameter or message roles for higher-level behavior instructions, and that instructions take priority over the prompt in the input parameter.
