Glossary · AI
Context window
Also called: context length, token limit, context size
A context window is the maximum number of tokens a language model can take into account in a single request, counting the instructions, the input data and the reply it writes.
Think of it as the model's working memory for one call. Everything it can "see" has to fit: the system prompt, the examples, the record you're asking about, any retrieved documents and the answer it's about to write. Anything beyond the limit is rejected or truncated, depending on the API.
What fills it faster than you expect
- Long email threads with quoted replies repeated in every message
- Whole JSON records with fields the step doesn't need
- Retrieved documents pasted in full instead of the relevant passages
- Conversation history in multi-turn agents, which grows every turn
Bigger isn't free
Modern models offer very large windows, but every token you send is billed and adds latency, and models can pay less attention to details buried in the middle of a very long input. A focused 2,000-token prompt usually beats a 100,000-token one for extraction and classification.
Working within it
Trim before you send: strip quoted replies, select fields, chunk long documents and summarize each chunk, or retrieve only the passages that matter (RAG). For agents, summarize older turns rather than carrying them verbatim.
In Fluxo Flow
Context limits come from the model you pick on your own provider account; Fluxo Flow doesn't change them. On the canvas, a Transformer node can map just the fields an AI step needs before it runs, a Loop can process long lists one item at a time, and the trace shows the tokens each step actually used so you can see where context is being spent.
Read next
Related terms
- TokenA token is the unit of text a language model reads and writes, usually a word, part of a word or a punctuation mark, and the unit AI providers use to price usage and set context limits.
- LLMAn LLM, or large language model, is a neural network trained on large amounts of text that generates text by repeatedly predicting the next token, which lets it summarize, classify, extract and draft.
- Retrieval-augmented generation (RAG)Retrieval-augmented generation (RAG) is a technique in which relevant information is looked up from your own data at request time and added to the prompt, so the model answers from those facts rather than from memory alone.
- Cron expressionA cron expression is a compact five-field string (minute, hour, day of month, month, day of week) that tells a scheduler exactly when to run a job.
- BYOK (bring your own key)BYOK, or bring your own key, means a software product runs its AI features on an API key from your own AI provider account, so usage is billed to you directly by the provider instead of being resold through the product.