Highlights:

  • Modular presents Mojo as a more practical option. Like Python, it has a reasonably simple syntax and can code a thousand times quicker.
  • The company claims that in addition to providing public access to its standard library code, it would also allow outside developers to submit enhancement requests.

Modular Inc. open sources Mojo’s core constituents, its programming language developed to code artificial intelligence software.

A modified version of the Apache 2 license is used to make the code public. Modular made changes to Mojo to make it simpler to integrate with programs released under the GPL, another well-liked open-source license. The Linux kernel is the most well-known GPL2 project.

Based in Palo Alto, California, Modular is supported by USD 130 million in funding received from Alphabet Inc.’s GV fund and some other investors. In August last year, the business released the Mojo programming language to the public. More than 175,000 developers and 50,000 organizations have used it since then.

Programming languages are frequently used in the writing of AI models. Developers implement the most basic parts of a neural network in Python, a language that is simple to learn but has comparatively sluggish performance. The remaining code is usually written in C , which is significantly more challenging to use and faster than Python.

Modular presents Mojo as a more practical option. Like Python, it has a reasonably simple syntax and can execute code a thousand times quicker. As a result, programmers don’t need to learn complicated languages like C to create quick AI models.

The so-called standard library of the language is made up of the Mojo components that Modular recently made open-sourced. This section of a programming language contains the fundamental syntactic components and other characteristics. Features like tweaking AI hyperparameters—the configurations that dictate how a neural network processes data—can be found in Mojo’s standard library.

The company claims that in addition to providing public access to its standard library code, it would also allow outside developers to submit enhancement requests. It will also regularly release updated versions of Mojo’s compiler. The intention is to enable developers to test their contributions to the standard library on the most recent compiler version.

Modular launched its commercial AI platform, MAX, at the end of the previous year, and it runs it on Mojo. The platform has capabilities to make tasks easier, like executing AI applications in environments with Kubernetes. The business revealed that it intends to eventually release a few of MAX’s components under an open-source license.

Simultaneously, a new platform version featuring multiple feature additions was released. One of the improvements, according to Modular, will make it simpler to combine Python and Mojo code in a single AI project. Additionally, the business has made performance enhancements aimed at accelerating AI inference.