While admitting that they “suck at naming their models,” OpenAI has launched GPT-4.1, along with GPT-4.1 Mini and GPT-4.1 Nano. These models are for developers and will not show up in your ChatGPT model picker. All three models are available now via API. According to OpenAI, the new models outperform GPT-4o in coding, instruction following, and long-context comprehension.
All three models support up to 1 million tokens — a sharp increase from GPT-4o’s 128,000-token limit — and include knowledge updates through June 2024. OpenAI reports GPT-4.1 scored 21% higher than GPT-4o and 27% higher than GPT-4.5 on coding benchmarks.
GPT-4.1 is 40% faster than GPT-4o, with input costs reduced by 80%. The model is 26% cheaper than GPT-4o and will replace GPT-4, which is being removed from ChatGPT on April 30. Access to GPT-4.5 via API will end in July.
The models deliver improved accuracy across long prompts and better code generation. Developers can analyze eight times more code at once, improving debugging and software prototyping. “Benchmarks are strong, but we focused on real-world utility, and developers seem very happy,” said OpenAI CEO Sam Altman.
Varun Mohan, CEO of Windsurf, joined the OpenAI livestream with an offer of free access to GPT-4.1 for seven days, starting yesterday. (I have no financial relationship with Windsurf, but as they say: “Free is very pro consumer.”) I spent a few hours with 4.1 using Windsurf last night – vibe-coding is absolutely the new-new thing.
Author’s note: This is not a sponsored post. I am the author of this article and it expresses my own opinions. I am not, nor is my company, receiving compensation for it. This work was created with the assistance of various generative AI models.