Ghostboard pixel Skip to content

Microsoft Launches GitHub Models To Help You Test AI Apps Easily

Microsoft's new attempt to help AI engineers.

Generative Artificial Intelligence (GenAI) has been the talk of the town for the past few years now. I know many of you are not big fans of it, but it has become a widely used piece of tech, with many people now using it to carry out various tasks.

One of the main things that drive it have been the many AI models, both open-source and closed-source, with OpenAI's GPT series being one of the most infamous ones out there.

Another organization that's been pushing for GenAI is Microsoft's GitHub, who introduced their Copilot code completion tool for developers back in 2021, and has not looked back since.

Now, similarly, they have introduced “GitHub Models”, a new AI-focused addition to their Marketplace, which is meant to make the lives of developers easier.

Let's check it out. 😃

GitHub Models: What to Expect?

From the get-go, GitHub Models is a new section for the Marketplace, where developers will find a large selection of AI models from different providers that can be tested in a built-in “Playground”.

It allows for testing prompts and model parameters for free, with the option for upgrading to a paid plan thanks to the tight Azure AI integration, with servers in over 25 Azure regions for some models.

During the announcement, GitHub added that:

We are launching GitHub Models, enabling our more than 100 million developers to become AI engineers and build with industry-leading AI models.

As for which models are available; at the time of writing there was an abundant list that included both open-source and proprietary options such as Mistral, Llama, Cohere, etc.

a screenshot of the various models on github models marketplace listing

There are a few from OpenAI too, such as GPT-4o, GPT-4o mini, and Microsoft's Phi-3 series of AI models.

If you click on one of those models, you will be shown a README, with the options to see the transparency, evaluation, license, and a few other important details, with the option to start a playground at the top.

a screenshot of a ai model on github model

GitHub mentions that this helps developers compare and test multiple AI models with ease, and there's also support for Codespaces, that allows for quick set up of a development environment.

They also clarified that no prompts or outputs are sent to model providers, and are not used to train or improve any models. If you are eager to learn more about GitHub Models, give the announcement a read.

Suggested Read 📖

14 Top Open Source LLMs For Research and Commercial Use
There are hundreds of open-source LLMs, here, we handpick some of the best ones for you to check out.

Get GitHub Models

You will have to join a waiting list if you want access to GitHub Models and the Playground, as they are being offered to only a handful of users as part of the limited beta.

But, you can see which models are available by visiting the GitHub Marketplace.

💬 Do let me know if you have already tried GitHub Models out!


More from It's FOSS...

Latest