# LLM (Large Language Model)

> An LLM (large language model) is an AI model trained on very large amounts of text that answers questions, summarises, translates and writes by predicting what text should come next. OpenAI GPT, Anthropic Claude and Google Gemini are well-known examples. Businesses use them to classify email, summarise documents and draft replies to customers.

URL: https://corrplus.net/glossary/llm/
Updated: 2026-10-22
Corrplus Team

**Definition:** An LLM (large language model) is an AI model trained on very large amounts of text that answers questions, summarises, translates and writes by predicting what text should come next.

## How does an LLM work?

An LLM splits text into small units called tokens and, at each step, predicts the most likely next token. Billions of parameters let those predictions capture grammar, context and general knowledge. However capable the model is, its knowledge stops at its training cut-off and it knows nothing about your internal information. That gap is usually closed with [RAG](/glossary/rag/).

## Where do businesses use LLMs?

In repetitive text work that rules cannot handle: routing incoming email by topic, extracting clauses from contracts, summarising meeting notes, drafting replies to customers. An LLM is rarely a product on its own. It creates the most value as a step inside a [workflow automation](/glossary/workflow-automation/) or an [AI agent](/glossary/ai-agent/).

## Choosing a model

Commercial models are reached through APIs from OpenAI, Anthropic, Google or Azure OpenAI. When data must not leave a given jurisdiction, open-weight models such as Llama, Mistral or Qwen can run on your own servers. For teams in Türkiye, models adapted to Turkish are also published on Hugging Face. The choice is shaped by data protection requirements (GDPR in the EU, [KVKK](/glossary/kvkk/) in Türkiye), language quality and cost. The tendency to produce confident but wrong answers is called [hallucination](/glossary/ai-hallucination/).

## Frequently asked questions

### Can LLMs handle Turkish and other non-English languages?

Leading commercial models generally read and write Turkish well. For legal or industry-specific terminology and for long documents, you should still test with your own examples before going live.

### Is it safe to send company data to an LLM provider?

That depends on the provider's enterprise terms, the region where data is processed and whether your data is used for training. For personal data, review international transfer rules with your legal adviser; where needed, an open-weight model can run on your own infrastructure.