AI
Large Language Model (LLM)
Also known as: LLM, Foundation model
A large language model is a neural network, typically transformer-based, trained on very large text corpora to predict the next token in a sequence, a capability that generalises to translation, summarisation, reasoning and code generation.
Capability scales with parameters, training data and compute, though post-training techniques such as instruction tuning and reinforcement learning from human feedback determine how usable a model actually is.
Practical deployment is governed less by raw capability than by cost per task, latency, context window and whether output quality can be measured reliably.