Open Source

Open source refers to software released under a license that grants anyone the right to view, modify, and redistribute the source code. The concept originated in the software development community as an alternative to proprietary, closed-source software, and is governed by licenses such as MIT, Apache 2.0, and GPL that define what users can and cannot do with the code.

In AI, the term has become more nuanced and sometimes contentious. Some model releases like Meta's Llama series and Mistral's models are described as open source because their weights are publicly downloadable, but they often come with usage restrictions that traditional open source licenses do not impose. The Open Source Initiative has argued that truly open source AI requires not just model weights but also training code, data, and documentation. The industry increasingly uses the term "open weight" to distinguish models that release weights but not training data or code from fully open source projects.

Open source has been a major driver of AI progress. Frameworks like PyTorch and TensorFlow, tools like Hugging Face Transformers and Ollama, and datasets on platforms like Hugging Face have democratized access to AI capabilities. Open source models allow researchers and companies to fine-tune, inspect, and deploy models without depending on API providers, enabling use cases where data privacy, cost control, or customization matter more than using the absolute frontier model.