How a language model works: tokens, context window and temperature
What a token is, what the context window limits, what temperature does and why language models are confidently wrong. Explained for developers, no maths.
Category
Artificial intelligence
Generative AI has changed how code is written, information is searched and tasks are automated. In this category we explain the concepts behind it (language models, tokens, context, agents) and how to use these tools with judgement.
We avoid empty promises. When a claim depends on a vendor announcement or documentation, we link to the original source, and we point out limitations as clearly as advantages.
- AI
- AI tools
- AI for developers
- Automation
- AI agents
Articles in Artificial intelligence
How to count the tokens in a prompt and estimate what a call to a language model costs
Why models charge per token, how to count tokens exactly for GPT and approximately for Claude, and how to cut the cost of each call without losing quality.
What an AI agent is and how it differs from a chatbot
An AI agent combines a language model with tools and a decision loop to complete tasks. How it differs from a chatbot, how it works, its risks and when to use one.