JetBrains Launches Mellum: An Open AI Model for Code Generation
JetBrains has launched Mellum, its first open AI model for coding, aimed at enhancing code generation and completion. With over 4 trillion tokens and 4 billion parameters, it supports various developer tools but requires fine-tuning before use. As AI-generated code becomes more prevalent, security challenges also arise, emphasizing the need for caution in its application.
USAGETOOLSWORKFUTURE
The AI Maker
1/22/20262 min read


JetBrains, known for its robust suite of app development tools, has taken an exciting step forward in the realm of artificial intelligence with the release of Mellum, its first open AI model specifically designed for coding. Available now on the AI development platform Hugging Face, Mellum aims to enhance the coding experience by providing intelligent code completion.
Trained on over 4 trillion tokens and boasting 4 billion parameters, Mellum is built to assist developers by generating code snippets based on the surrounding context. The parameters of an AI model can be thought of as its problem-solving prowess, while the tokens represent the raw data the model uses to learn and generate outputs. To put it in perspective, a million tokens is roughly equivalent to 30,000 lines of code, showcasing the extensive training Mellum has undergone.
JetBrains highlights that Mellum is designed for integration into various professional developer tools, such as integrated development environments (IDEs) and AI-powered coding assistants. It also holds potential for educational applications and fine-tuning experiments, making it a versatile tool for both seasoned developers and those just starting out.
However, it’s important to note that Mellum requires some effort to get up and running. The base model isn’t ready to use straight out of the box; it needs to be fine-tuned first. JetBrains has provided a few pre-fine-tuned models for Python, but these are primarily for estimating potential capabilities rather than being production-ready.
As AI continues to transform the software development landscape, it also brings new security challenges to the forefront. According to a recent survey by Snyk, over 50% of organizations have encountered security issues with AI-generated code, whether sometimes or frequently. This highlights the necessity for caution when deploying AI tools in real-world applications.
JetBrains also reminds users that Mellum may reflect biases found in public codebases, which can lead to generated code that mirrors styles from open-source repositories. Consequently, the code suggestions produced by Mellum may not always be secure or free from vulnerabilities.
In a blog post, JetBrains expressed that this is merely the beginning of their journey with Mellum. They’re not aiming for broad generality but rather focusing on practical applications. The company hopes that even a single meaningful experiment or contribution stemming from Mellum will be viewed as a success. As the world of AI-generated code continues to evolve, tools like Mellum will play a crucial role in shaping the future of software development.
Cited: https://techcrunch.com/2025/04/30/jetbrains-releases-mellum-an-open-ai-coding-model/
Your Data, Your Insights
Unlock the power of your data effortlessly. Update it continuously. Automatically.
Answers
Sign up NOW
info at aimaker.com
© 2024. All rights reserved. Terms and Conditions | Privacy Policy
