Independent AI reporting · model releases · research · culture of intelligenceUpdated November 19, 2026
← All stories
AI BASICS / INFORMATIONAL

What Is a Large Language Model (LLM)?

A clear introduction to tokens, training, inference, context and common LLM limitations.

What Is a Large Language Model (LLM)? Original editorial illustration · Runway Presstoday

Original editorial illustration for this topic. Not an official product screenshot or a depiction of the exact reported event.

Independent educational guide. Features, model versions, pricing and interfaces may change. Check the linked official documentation before purchase or deployment. Examples are illustrative, not screenshots of a tested product.

A large language model is a system trained on text and often other types of data to predict and generate sequences. During inference it produces output conditioned on the current input and available context. Modern systems can answer questions, draft code and transform documents, but fluent language should not be confused with an independent guarantee of factual truth.

What to evaluate before choosing a tool

An LLM uses tokenization to break input into units and relies on learned statistical patterns. Training develops the model parameters; a normal chat message is not the same process as retraining the base model. Some applications combine a model with external search, tools or retrieval so that it can use recent or organization-specific information.

Step-by-step workflow

  1. Start with a narrow task, such as rewriting a known paragraph without adding facts.
  2. Supply the audience, desired format and source material.
  3. Ask the model to identify unsupported claims or assumptions.
  4. Verify dates, figures and quotes against sources before publication.
  5. Measure success by the quality of the actual task, not by how human the response sounds.

Practical example

Here is an original example you can adapt to your own project. Its wording is a starting point, not a guarantee of a particular result.

Summarize the three paragraphs below for a nontechnical audience. Use 100–130 words. Do not introduce claims beyond the supplied text; mark any unresolved question clearly.

Common mistakes and limitations

Language models can hallucinate sources, misunderstand ambiguous instructions and carry old information. They also have context and output limits. For sensitive work, use human review and trustworthy external evidence.

When this approach makes sense

Think of an LLM as a flexible text-and-reasoning component in a larger workflow. The surrounding data, validation, permissions and evaluation matter as much as the model itself.

Frequently asked questions

Is an LLM a search engine?

No. A standalone model generates from its learned patterns and context; search can be integrated separately.

Does a chat message train the model immediately?

Not in the sense of updating the deployed base model weights during that conversation.

Official sources and documentation

These primary references describe the products and concepts used in this guide. They are not evidence that Runway Presstoday independently tested every feature.

Continue learning

Browse all AI Basics articles ↗