If you have been working with Python for AI or machine learning, you already know the trade-off: Python is easy to write, but it can be painfully slow when performance really matters. That is exactly the gap Modular set out to close with Mojo.<\/p>\n\n\n\n
Mojo is not just another programming language. It is designed from the ground up for AI development, borrowing Python’s readable syntax while delivering performance that rivals C++ and Rust. For developers who work on model training, inference pipelines, or hardware-level AI systems, that combination is significant.<\/p>\n\n\n\n
In this guide, you will get a complete introduction to Mojo. You will understand why it was built, how it works, and how to start writing code in it, even if your background is primarily in Python.<\/p>\n\n\n\n
TL;DR Summary<\/strong><\/p>\n\n\n\n Mojo is a programming language developed by Modular, the company founded by Chris Lattner, the original creator of Swift and a key contributor to LLVM and Clang.<\/p>\n\n\n\n It is designed to be a superset of Python<\/a>, which means valid Python code is also valid Mojo code in most cases. But Mojo adds features that Python simply cannot offer, things like static typing, manual memory control, and low-level hardware access.<\/p>\n\n\n\n The goal is straightforward: give AI and ML<\/a> developers a language that feels familiar but performs at a level that Python never could.<\/p>\n\n\n\n Think of Mojo as Python that has been rebuilt with a systems programming foundation underneath it.<\/p>\n\n\n\n To understand why Mojo exists, it helps to understand the problem it is solving.<\/p>\n\n\n\n Python is the dominant language in AI and data science, but its performance ceiling is a real limitation. Most production AI systems deal with this by writing performance-critical code in C++<\/a> or CUDA and wrapping it with Python. It works, but it creates a fragmented development experience; you are essentially maintaining two codebases in two very different languages.<\/p>\n\n\n\n Mojo was created to eliminate that split.<\/p>\n\n\n\n With Mojo, you can write everything, from high-level data processing logic to low-level hardware kernels, in a single language. That is a meaningful shift for teams building large-scale AI systems.<\/p>\n\n\n\n There are a few reasons why this matters right now:<\/p>\n\n\n\n Mojo addresses all of these directly.<\/p>\n\n\n\n Getting Mojo running on your machine is straightforward. Here is how to do it step by step.<\/p>\n\n\n\n If you want to try Mojo without installing anything locally, Modular offers a browser-based playground. Head to developer.modular.com<\/a> and sign up for access. This is the fastest way to get started and is ideal if you just want to explore the syntax.<\/p>\n\n\n\n For local development, you will need to install the Modular CLI first.<\/p>\n\n\n\n Step 1: Install the Modular CLI:<\/strong><\/p>\n\n\n\n Step 2: Authenticate your account:<\/strong><\/p>\n\n\n\n This will prompt you to log in with your Modular account. Create one at developer.modular.com if you have not already.<\/p>\n\n\n\n Step 3: Install Mojo:<\/strong><\/p>\n\n\n\n Step 4: Verify the installation:<\/strong><\/p>\n\n\n\n If you see a version number printed, you are ready to go.<\/p>\n\n\n\n Modular offers an official Mojo extension for Visual Studio Code. Search for “Mojo” in the VS Code extensions marketplace and install the one published by Modular. It gives you syntax highlighting, code completion, and inline error checking as you write.<\/p>\n\n\n\n Once your environment is set up, create a new file with a .mojo extension.<\/p>\n\n\n\n Run it from your terminal:<\/p>\n\n\n\n You should see:<\/p>\n\n\n\n That is your first Mojo program. Notice how similar it looks to Python, the structure is intentionally familiar. The key difference is the use of fn instead of def, which we will cover in detail shortly.<\/p>\n\n\n\n Mojo introduces several features that Python does not have. These are not optional extras, they are central to how Mojo achieves its performance. Here is what you need to understand early on.<\/p>\n\n\n\n In Python, you can assign any value to any variable at any time. Mojo lets you declare variable types explicitly, which helps the compiler optimise your code.<\/p>\n\n\n\n Using let declares an immutable variable \u2014 one that cannot be reassigned after initialisation. For mutable variables, use var:<\/p>\n\n\n\n This distinction matters. Immutability allows the compiler to make stronger guarantees about your code’s behaviour, which translates into better optimisations.<\/p>\n\n\n\n SIMD stands for Single Instruction, Multiple Data. It is a hardware feature that lets you perform the same operation on multiple data points simultaneously \u2014 critical for numerical computing and AI workloads.<\/p>\n\n\n\n Mojo exposes SIMD directly in the language:<\/p>\n\n\n\n This kind of low-level hardware access is something Python simply cannot offer without external libraries written in C.<\/p>\n\n\n\n Mojo introduces structs, which are similar to Python classes but are resolved at compile time rather than runtime. This makes them significantly faster for performance-critical use cases.<\/p>\n\n\n\n Structs in Mojo are value types by default, they are copied, not referenced, which eliminates a common class of bugs in concurrent code.<\/p>\n\n\n\n\n
What is Mojo?<\/strong><\/h2>\n\n\n\n
Why Was Mojo Created?<\/strong><\/h2>\n\n\n\n
\n
Setting Up Your Mojo Environment<\/strong><\/h2>\n\n\n\n
Option 1: Use the Mojo Playground (Easiest)<\/strong><\/h3>\n\n\n\n
Option 2: Install Mojo Locally<\/strong><\/h3>\n\n\n\n
curl https:\/\/get.modular.com | sh -<\/code><\/p>\n\n\n\nmodular auth<\/code><\/p>\n\n\n\nmodular install mojo<\/code><\/p>\n\n\n\nmojo --version<\/code><\/p>\n\n\n\nSetting Up VS Code for Mojo<\/strong><\/h3>\n\n\n\n
Writing Your First Mojo Program<\/strong><\/h2>\n\n\n\n
fn main():\n\n print(\"Hello from Mojo!\")<\/code><\/pre>\n\n\n\nmojo hello.mojo<\/code><\/p>\n\n\n\nHello from Mojo!<\/code><\/p>\n\n\n\nCore Features of Mojo You Should Know<\/strong><\/h2>\n\n\n\n
Static Typing<\/strong><\/h3>\n\n\n\n
let name: String = \"Mojo\"\n\nlet version: Int = 1<\/code><\/pre>\n\n\n\nvar counter: Int = 0\n\ncounter = counter + 1<\/code><\/pre>\n\n\n\nSIMD Support<\/strong><\/h3>\n\n\n\n
from DType import float32\n\nlet vec = SIMD[float32, 4](1.0, 2.0, 3.0, 4.0)<\/code><\/pre>\n\n\n\nStructs<\/strong><\/h3>\n\n\n\n
struct Point:\n\n var x: Float32\n\n var y: Float32\n\n fn __init__(inout self, x: Float32, y: Float32):\n\n self.x = x\n\n self.y = y<\/code><\/pre>\n\n\n\n