Mehmet EmrahKonyaTR
← Back to glossary

Large language model

Büyük dil modeli

D2

A large language model is a deep-learning language model trained on a very large text dataset; a generative LLM commonly produces text through next-token or next-word prediction.

Review status: 2026-11-26

Technical explanation

A generative LLM can produce text by predicting the next token or word in a sentence or phrase.

Conceptual boundaries

An LLM is not a Transformer architecture itself; in its March 2025 publication, NIST described Transformer-based GPT models as then-predominant LLM architecture.

Provider-neutral example

A helpdesk workflow can use an LLM to propose the next sentence of a response for an employee to edit before sending.

Limitations

Next-token or next-word prediction is a common generative mechanism and does not define every language-model architecture.

Related concepts

Atomic claims and evidence

  1. 1.1NIST SP 1500-29, Artificial Intelligence in the Fire Service: Considerations for Implementing Artificial Intelligence into Electronic Safety Equipment
    Source
    NIST SP 1500-29, Artificial Intelligence in the Fire Service: Considerations for Implementing Artificial Intelligence into Electronic Safety Equipment
    Source role
    Authoritative source
    Exact locator
    Section 1.1, p. 5: Large Language Models (LLM) bullet; sentences distinguishing Generative LLMs from Discriminatory LLMs
    Supported claim
    NIST SP 1500-29 defines a large language model as a deep-learning language model trained on a very large text dataset and distinguishes generative from discriminatory LLMs; in its generative-AI risk discussion, NIST AI 600-1 states that LLMs predict the next token or word in a sentence or phrase.
    Last verification
    Review due
    Scope limitation
    This distinguishes a broad LLM category from generative LLMs and does not set a parameter-count threshold.
    1.2NIST AI 600-1, Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile
    Source
    NIST AI 600-1, Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile
    Source role
    Authoritative source
    Exact locator
    Confabulation pp. 5-6: LLMs predict the next token or word in a sentence or phrase
    Supported claim
    NIST SP 1500-29 defines a large language model as a deep-learning language model trained on a very large text dataset and distinguishes generative from discriminatory LLMs; in its generative-AI risk discussion, NIST AI 600-1 states that LLMs predict the next token or word in a sentence or phrase.
    Last verification
    Review due
    Scope limitation
    This distinguishes a broad LLM category from generative LLMs and does not set a parameter-count threshold.
  2. 2.1NIST AI 100-2e2025, Adversarial Machine Learning: A Taxonomy and Terminology of Attacks and Mitigations
    Source
    NIST AI 100-2e2025, Adversarial Machine Learning: A Taxonomy and Terminology of Attacks and Mitigations
    Source role
    Authoritative source
    Exact locator
    Appendix A p. 108: generative pre-trained transformer (GPT)
    Supported claim
    In its March 2025 publication, NIST described GPT as a Transformer-based model family and as the then-predominant architecture for large language models.
    Last verification
    Review due
    Scope limitation
    Predominant does not mean universal, and the claim does not make an LLM identical to its architecture.