LLM (Large Language Model)
An LLM is an AI model trained on large volumes of text that can read, generate and reason about natural language — used in production systems for tasks like structured data extraction, classification and drafting text.
Large language models predict and generate text based on patterns learned from vast amounts of training data. In a production engineering context, the useful capability isn't chat — it's using an LLM as a component inside a pipeline: reading an unstructured document and extracting specific fields, classifying a support ticket, or drafting a response for a human to review.
The gap between an LLM demo and a production LLM pipeline is reliability. A single prompt-and-parse call can hallucinate a field that isn't in the source text, return malformed output, or cost more than necessary by routing every request through the most expensive model available regardless of complexity. Production pipelines add schema validation, source traceability so an extracted value can be checked against the original passage, and cost-aware model routing.
LLM cost is also a real engineering variable, not an afterthought — batch processing (trading latency for lower per-token cost), caching, and choosing a cheaper model for simpler sub-tasks are standard levers. See our AI API Cost Calculator for estimating what a given volume and model choice would actually cost to run monthly.