Earlier today, OpenAI announced GPT-4.1 models targeted at developers. The GPT-4.1 model family includes GPT-4.1, GPT-4.1 mini, and GPT-4.1 nano, and these are available only via APIs.
According to OpenAI, these new purpose-built models for developers deliver significant improvements in coding, instruction following, and long-context understanding. The OpenAI team has optimized GPT-4.1 for real-world coding scenarios, including frontend coding, following formats reliably, adhering to response structure and ordering, and more.
For the first time, OpenAI is supporting up to 1 million tokens of context with these models. They also feature a refreshed knowledge cutoff of June 2024.
We’re bringing the latest GPT-4.1 models from OpenAI to Foundry, representing significant advancements in coding, instruction following, and long-context processing. Big upgrades for devs building with AI. https://t.co/uTNdUg2Izr
— Satya Nadella (@satyanadella) April 14, 2025
Today, Microsoft announced that the GPT-4.1, 4.1-mini, and 4.1-nano models are now available on Microsoft Azure OpenAI Service and GitHub. To address the needs of enterprise customers, Microsoft will also be adding support for supervised fine-tuning for GPT-4.1 and 4.1-mini later this week.
This will allow enterprises to customize the base models based on their own datasets, with specific tone, domain terminology, and task workflows. Additionally, the fine-tuned models will be managed and deployed through the Azure AI Foundry.
On GitHub, the new GPT-4.1 is available in GitHub Copilot and GitHub Models. Copilot Enterprise administrators will need to enable access to this new GPT-4.1 model for their users through a new policy in Copilot settings. GitHub highlighted that the new GPT-4.1 model will perform better than the previous workhorse model, GPT-4o, across the board, with major improvements in coding.
This new model is now available for all GitHub Copilot users and can be accessed via the model picker under the name GPT-4.1 (Preview) in Visual Studio Code and on the github.com chat. Developers can also try GPT-4.1 alongside other models in the GitHub Models playground.
2 Comments - Add comment