What Is a Large Language Model (LLM)?

A large language model (LLM) is a type of artificial intelligence trained on billions of words of text to understand, generate, and reason about human language. Models like GPT, Claude, Gemini, and Qwen are all LLMs.

How Do LLMs Work?

LLMs are built on the transformer architecture. During training, the model learns statistical patterns across massive text corpora — books, websites, code, and more. This allows it to:

• Generate coherent, contextually relevant text
• Summarize long documents
• Translate between languages
• Write code in dozens of programming languages
• Reason through multi-step problems

Key Concepts

• Parameters: The learned weights that define the model’s knowledge. Frontier models are estimated to range from hundreds of billions to over a trillion parameters.
• Tokens: LLMs process text as tokens — subword units that the model reads and generates.
• Context window: The maximum amount of text an LLM can process in a single conversation.
• Temperature: A setting that controls how creative or deterministic the model’s output is.

Popular LLMs

Model	Provider	Key Strength
GPT-6 Astra	OpenAI	General-purpose reasoning
Claude Sonnet 5	Anthropic	Long context, coding, safety
Gemini 3.8 Flash	Google	Multimodal, long context
Muse Spark	Meta	Long-horizon agentic work
Qwen3	Alibaba	Open-weight, local deployment

Running LLMs Locally

With tools like Ollama, you can run open-source LLMs directly on your Mac without sending data to the cloud. Elvean connects to both local models via Ollama and cloud APIs — giving you the best of both worlds in one native app.