Open source LLMs
Definition
Open source LLMs are large language models whose weights are publicly available for download, allowing anyone to self-host, fine-tune, and inspect them. Families such as Llama, Mistral, Qwen, and DeepSeek give organizations control over deployment, customization, and data privacy, in contrast to closed models accessible only through a provider's API.
How it works
An open source LLM ships with downloadable weights, the trained parameters that define the model. With the weights in hand, you can run the model on your own infrastructure using serving frameworks, quantize it to fit smaller hardware, or fine-tune it on proprietary data.
Licenses vary. Some releases are fully permissive, while others restrict commercial use or large-scale deployment, so the term open is often more precisely open-weight. Training data and code are not always released even when weights are.
Popular families include Llama, Mistral, Qwen, and DeepSeek. These models can be paired with retrieval-augmented generation, agent frameworks, and tool calling just like closed models, and a broad ecosystem of tooling exists for hosting and optimizing them.
Why it matters
Open source LLMs give organizations control they cannot get from a closed API. Self-hosting keeps sensitive data inside their own environment, which is critical for regulated industries and strict data-privacy requirements. Owning the deployment also removes dependence on a single vendor's pricing and availability.
Customization is the other draw. Teams can fine-tune open models on domain data to outperform general-purpose models on narrow tasks, and inspect weights for transparency. The trade-off is operational: you take on the cost and engineering of running and maintaining the model yourself.
Frequently asked questions
Are open source LLMs as good as closed models?
The gap has narrowed considerably. Leading open models are competitive with closed models on many benchmarks, though the very top frontier models are often still closed. For most production tasks, a well-chosen open model is more than capable.
What does open weight mean versus open source?
Open weight means the trained parameters are downloadable, while open source traditionally implies the training code and data are also available. Many models release only weights, sometimes under licenses that restrict certain uses, so open weight is the more precise term.
Why self-host an open source LLM?
Self-hosting keeps data inside your own infrastructure for privacy and compliance, removes per-token API fees at scale, and avoids vendor lock-in. The trade-off is the engineering effort to deploy, scale, and maintain the model reliably.
Can I fine-tune an open source LLM on my own data?
Yes. Access to the weights lets you fine-tune on proprietary or domain-specific data, often producing a model that outperforms larger general-purpose models on your specific tasks while keeping your data private.
Large language model (LLM)
A large language model is an AI system trained on vast amounts of text to understand and generate human language. Built on transformer architecture and containing billions of parameters, LLMs predict the next token in a sequence, enabling them to answer questions, write, summarize, and reason. They power modern chat assistants, AI search, and autonomous agents.
Foundation models
Foundation models are large-scale AI models trained on broad, diverse data that serve as a general-purpose base adapted for many downstream applications. Rather than building a model per task, organizations fine-tune or prompt a single foundation model for translation, summarization, coding, search, and more. Large language models and multimodal models are common examples.
Small language models (SLMs)
Small language models are compact AI models, typically ranging from about one to ten billion parameters, designed for on-device deployment, low latency, and cost efficiency while retaining useful capability. By trading some breadth for a smaller footprint, SLMs run on phones, laptops, and edge hardware, enabling private, fast, and inexpensive language tasks.
DeepSeek
DeepSeek is a Chinese AI lab known for the DeepSeek V3 model and the R1 reasoning model. Its models use a mixture-of-experts architecture, are released under permissive MIT licensing, and are positioned as competitive with frontier systems while emphasizing strong performance at lower training and inference cost.
AI fine-tuning
AI fine-tuning is the process of taking a pre-trained model and training it further on a smaller, specialized dataset so it adapts to a specific task, domain, tone, or format. It adjusts the model's existing weights rather than training from scratch, producing outputs that better match a brand's requirements or a narrow use case at lower cost than full training.
Data privacy in AI
Data privacy in AI covers the practices that protect personal and sensitive information across the AI lifecycle — what enters training data, what is sent through APIs, how enterprise deployments isolate data, and how systems meet regulations like GDPR. It addresses consent, retention, data residency, and whether user inputs are used to further train models.