Large Language Model (LLM)
A neural network trained on massive text corpora to understand and generate human-like language.
A Large Language Model is a transformer-based neural network with billions of parameters, trained on trillions of tokens of text to predict the next token in a sequence. This simple objective, applied at scale, produces emergent capabilities including reasoning, summarization, translation, and code generation.
Modern LLMs like GPT-4, Claude, Gemini, and Llama power everything from chat assistants to autonomous agents. Enterprises typically consume them via API rather than training from scratch — the compute cost of pre-training a frontier model exceeds USD 100M.
For production use, LLMs are combined with retrieval (RAG), tools, and guardrails to ground responses in company data and prevent hallucinations.
Key points
- Trained on public web, books, and code — not on your private data by default
- Context window (typically 8K–2M tokens) limits how much text can be processed at once
- Cost scales with input + output tokens; smaller models are 10–100x cheaper
- Fine-tuning or RAG is required to specialize on domain knowledge
Common use cases
Frequently asked
Related terms
A pattern that grounds LLM answers in your private documents by retrieving relevant snippets at query time.
Continuing training of a pre-trained model on domain-specific data to specialize its behavior.
The practice of designing model inputs to produce reliable, task-specific outputs.
An LLM-driven system that plans, calls tools, and executes multi-step tasks autonomously.
Applying Large Language Model (LLM) in your business?
We help enterprises design, ship, and govern AI systems end-to-end.