GPT-4.1 nano
OpenAI · United States · 2025
Ten cents per million input tokens against a million-token window — and the only member of the GPT-4.1 family with a shutdown date.
GPT-4.1 nano is the entry tier of the GPT-4.1 family, released on 14 April 2025. It is the cheapest way OpenAI ever offered to read a very large prompt: $0.1 per million input tokens, $0.025 cached, $0.4 per million output — against the same 1,047,576-token window as the flagship of the family. The rest of the specification follows the generation: 32,768 tokens of output, text and image input, no reasoning step, and a knowledge cutoff of 1 June 2024. It was designed for classification, autocompletion and other high-volume work where latency and unit cost decide the architecture. On 22 April 2026 OpenAI put it on a list of legacy models to be shut down on 23 October 2026, naming GPT-5.6 Luna as the replacement. That makes nano the first casualty of the family: the base model and the mini version were left off the list and are still served.
▸Videos
No videos yet.
▸Reviews
No reviews yet. Be the first!