> Term
Context Window
The maximum amount of text or tokens an LLM can process in a single request.
Detailed Explanation
The context window dictates how much recent memory or retrieved knowledge an AI agent can hold before it forgets the beginning of the conversation.
Why It Matters
Exceeding the context window either truncates crucial information or crashes the API request entirely.
Common Failure Mode
Practical Example
Production Manifestation
Token counters, prompt truncation logic, and sliding window memory handlers.
Frequently Asked Questions
What is Context Window in short?
The maximum amount of text or tokens an LLM can process in a single request.
What is the most common failure mode?
Stuffing an entire codebase into a prompt, overflowing the context window, and causing a 400 Bad Request error.
AI Summary
The maximum amount of text or tokens an LLM can process in a single request. Exceeding the context window either truncates crucial information or crashes the API request entirely.