Key takeaways
- In AI, LLM stands for large language model: a neural network trained on huge amounts of text to understand and generate language.
- In law, LL.M. is a Master of Laws, a postgraduate law degree.
- Examples of LLMs include GPT, Claude, Gemini, Llama, Mistral, Qwen and DeepSeek.
- A local LLM runs on your own hardware; inference is running a trained model to produce output.
- You can build a tiny LLM to learn, but real products adapt existing models with prompting, retrieval or fine-tuning.
LLM has two common meanings. In AI, it stands for large language model: a neural network, usually a transformer, trained on huge amounts of text to understand and generate language, powering tools such as ChatGPT, Claude and Gemini. In law, LL.M. stands for Master of Laws, a postgraduate degree. This page covers both meanings, then focuses on AI: examples, why we use LLMs, what "LLM model" means, local LLMs, inference, how to make an LLM, and the best courses, drawing on my work building RAG and fine-tuning systems.
What does LLM mean?
In technology, LLM means large language model: an AI model trained on vast text data that can answer questions, write, summarise, translate, reason and code. "Large" refers to the billions of parameters and the scale of training data.
Also asked as: what is llm means · what is llm stands for · what is llm examples
LLMs work by predicting the next token, a word piece, repeatedly, which turns out to capture grammar, facts, reasoning patterns and style. They are the core of modern AI assistants and many software features. In other contexts, LLM can mean something else, most commonly LL.M., the law degree, so the meaning depends on whether you are reading about technology or law.
What does LLM stand for?
In AI, LLM stands for Large Language Model. In law and academia, LL.M. stands for Legum Magister, Latin for Master of Laws.
The AI term became widespread after 2020, with models like GPT-3, and especially after ChatGPT launched in late 2022. You will see it in phrases such as "LLM-powered app", "LLM API" and "open-source LLM". Related terms include SLM, small language model, for compact models that run on devices, and multimodal model, for models that also handle images, audio or video.
What does LLM mean in law?
In law, LL.M., Master of Laws, is a postgraduate law degree, usually taken after a first law degree such as an LL.B. or the five-year integrated law degree in India. It allows specialisation in areas such as corporate law, international law, intellectual property or human rights.
Also asked as: what is llm means in law
In India, a one-year LL.M. is offered by many universities and national law universities, with admission often through entrance tests such as CLAT for national law universities. Abroad, LL.M. programmes are popular with lawyers seeking international exposure. Interestingly, the AI and law meanings now meet: lawyers increasingly use large language models for research and drafting, with careful verification, since AI can invent case citations.
What are examples of LLMs?
Examples include OpenAI's GPT models behind ChatGPT, Anthropic's Claude, Google's Gemini, Meta's Llama, Mistral's models, Alibaba's Qwen, DeepSeek's models and Microsoft's Phi.
They fall into two groups. Proprietary models such as GPT, Claude and Gemini are used through apps and APIs. Open-weight models such as Llama, Qwen, Mistral, Gemma and DeepSeek can be downloaded and run locally or on your own servers. There are also specialised LLMs for coding, embeddings for search, and domain-tuned models for areas such as medicine or law. Model versions change quickly; families are more stable than version numbers.
Why do we use LLMs?
We use LLMs because they handle language tasks flexibly and quickly: answering questions, drafting and editing text, summarising long documents, translating, writing and explaining code, extracting structured data, and acting as the reasoning engine in AI agents.
Their main advantage is generality: one model handles many tasks from instructions, without task-specific training. Their main weakness is reliability: they can be confidently wrong. So production systems pair LLMs with retrieval for facts, validation for structured outputs, tools for precise computation, and human review for high-stakes decisions. RAG.NextUpgrad, for example, grounds answers in retrieved documents with citations and refuses when evidence is weak.
What does "LLM model" mean?
"LLM model" is a common, slightly redundant phrase meaning a specific large language model, such as a particular version of Llama or Claude. Strictly, LLM already includes "model", but the phrase is widely used.
Also asked as: what is llm model means
It is often used to distinguish the model from the app around it: ChatGPT is an app; the GPT model inside it is the LLM. When someone asks "which LLM model should I use?", they are asking which specific model and version fits their task, cost and privacy needs. Answer it with a small evaluation on your own examples rather than by reputation.
What does local LLM mean?
A local LLM is a language model that runs on your own computer or server instead of a cloud service. Your prompts and data stay on your hardware, and there is no per-request cost.
Also asked as: what is local llm means
Tools such as Ollama, llama.cpp and LM Studio make running local LLMs easy, using quantised open-weight models that fit in laptop memory. Local LLMs are great for privacy, offline use, experimentation and high-volume simple tasks; they are weaker than frontier cloud models on hard reasoning and complex coding, and limited by your hardware. A laptop with 16 GB of RAM runs 7B to 8B models comfortably at 4-bit quantisation.
What does LLM inference mean?
LLM inference means using a trained model to produce output: feeding it a prompt and generating a response, token by token. It is distinct from training, which adjusts the model's parameters.
Also asked as: what is llm inference means
Inference cost and speed depend on model size, input and output length, hardware and serving software. Key metrics are latency to the first token, tokens per second and cost per request. Techniques such as quantisation, caching of repeated prompt prefixes, batching and efficient serving engines like vLLM improve inference efficiency. When you pay for an LLM API, you are paying for inference, usually per token of input and output.
How do I make an LLM?
To learn, build a tiny language model from scratch: a small transformer trained on a text dataset with PyTorch, following educational projects such as nanoGPT. To build products, adapt existing models rather than training from scratch, which needs enormous data and compute.
Also asked as: how to make llm
Three practical levels:
- Learn: implement a small GPT on a dataset of a few megabytes to understand tokens, attention and training.
- Adapt: fine-tune an open model with LoRA or QLoRA on your task, as FineTune Studio does from the browser.
- Build products: combine an existing model with retrieval, tools and evaluation.
Training a frontier-scale model is done by well-funded labs; almost all practical work happens at the adapt and build levels.
Which LLM course is best?
Good options include Hugging Face's free LLM course for practical skills, Andrej Karpathy's videos on building GPT from scratch for deep understanding, DeepLearning.AI's short courses on building with LLMs, and the official documentation and cookbooks from model providers.
Choose by goal:
- Understand how LLMs work: Karpathy's build-from-scratch series and the Transformer paper.
- Use open models: Hugging Face's course.
- Build applications: short courses on RAG, agents and evaluation, plus provider docs.
- Fine-tuning: Hugging Face PEFT and TRL documentation with a small project.
The best course is one you finish with a project: for example, a document Q&A app with citations and an evaluation set.
Sources
GenAI engineer and AI product builder with 2+ years shipping production-grade AI systems: RAG pipelines, fine-tuned LLMs, hybrid retrieval and multi-modal apps across vision, speech and OCR, architected end to end from ingestion to deployment. Leads engineering for SCULT INDIA's 14-member team, founded the 500+ member TechVerse Enclave community and has mentored 200+ students. Three hackathon first prizes: Changethon 2025 (IIT Roorkee), Product Genesis at Vividhotsava 2025 (CSJMU) and BYTEBATTLE (MeetKats).
Open to GenAI roles, hackathon judging, mentorship sessions and guest talks at colleges. For campus sessions the inviting institution covers the visit end to end: travel (a First AC train ticket for journeys up to five hours, a flight for anything longer), accommodation, meals, and pickup and drop; the honorarium and workshop fees are negotiable on WhatsApp or on the call, and requests are welcome from institutions that can cover the visit this way. Email pranjulrathour41@gmail.com.
