Context Window

LLM’s short-term memory: Max tokens it can process at once.

From 10-AI-Concepts

• Early: 4k tokens (GPT-3).
• Now: >1M.
• Impacts prompt design, conversation memory.