Large Language Model (LLM)
A neural network trained on very large amounts of text that predicts the next token, which lets it generate, summarise, classify and reason over language.
Definition
A large language model is a transformer-based neural network trained on a very large body of text. During training it learns one thing extremely well: given a sequence of tokens, predict the next one. That one objective, trained on enormous volumes of text, produces a model that can draft, translate, summarise, extract, classify and follow instructions in natural language. Claude, GPT, Gemini, Llama and Mistral are all LLMs.
"Large" refers to the number of parameters (the learned weights) and the volume of training data. After pre-training, most commercial models go through further stages, such as instruction tuning and RLHF, so they behave like a helpful assistant rather than a text-completion engine.
Example: a customer emails "my invoice says £1,240 but the quote was £1,100". An LLM can read the message, classify it as a billing dispute, extract both figures and draft a reply. It does this with no billing-specific training, because the pattern is already present in the language it learnt from.
What it means in practice
- One general model replaces many narrow ones: the same LLM can handle support, document extraction and internal search.
- You pay per token, so cost rises with usage rather than with engineering effort.
- Models are probabilistic. Without evals and guardrails, quality and safety drift.
- Choice of model is reversible. Good architecture lets you swap providers as prices and model quality change.
How Teksmobile uses it
LLMs sit at the centre of our LLM application development work, where we build on Claude, OpenAI, Gemini and open-weight models depending on the client's cost, latency and data-residency needs. Our AI strategy engagements help you decide which use cases an LLM should own and which it should not.
Related terms
Part of the Teksmobile AI Glossary. For longer reads, see our guides.
Frequently asked questions
Is an LLM the same as ChatGPT?
No. ChatGPT is a product built on top of an LLM. The model is the engine; the chat interface, memory, tools and safety layers are the application around it. Your business application will have the same shape, with your own data and rules.
Which LLM should we use?
It depends on the task, the data rules and the budget. We usually start with a hosted frontier model (Claude, GPT or Gemini) to prove the use case, then evaluate smaller or open-weight models where cost or data residency matters. See on-prem vs cloud LLMs.
Can an LLM access our company data?
Only if you give it that data. By default a model knows nothing about your systems. Retrieval-augmented generation (RAG) and tool use are the standard ways to connect a model to your documents and applications safely.