American AI company OpenAI has recently revealed a new series of AI models – GPT 4.1, GPT-4.1 mini, and GPT-nao. The latest models are reported to exceed the performance of previous GPT models, particularly in coding and instruction following. The newer models also offer up to 1 million token input, which is substantially more than the 128K context window that GPT-4o provides.
“Today, we’re launching three new models in the API: GPT‑4.1, GPT‑4.1 mini, and GPT‑4.1 nano. These models outperform GPT‑4o and GPT‑4o mini across the board, with major gains in coding and instruction following”, the company’s official blog states.
On several benchmarks, the GPT-4.1 model outperforms GPT-4o, including in long context processing. The GPT 4.1 is more adept at handling complex coding problems and generates cleaner front-end code.
See Related: OpenAI Releases GPT-4o: A Faster And More Capable Version of GPT- 4
The GPT-4.1 model series is endorsed as a more cost-effective option for superior results. The GPT 4.1 will cost $2.0 per 1 million input tokens compared to GPT-4o’s $5.0 and $8 per million output tokens compared to $20 in GPT-4o. Meanwhile, GPT-4.1 nano is the company’s cheapest model that still offers high performance.
OpenAI has also announced that fine-tuning will be available for the GPT-4.1 series of models very soon. This will allow developers to adjust the specifics of the models to suit specific requirements. Currently, the models are available in the OpenAI API and Microsoft’s Azure OpenAI service.