Definition: The maximum amount of information that a model can have available in a single request.

Simply put: It is like the model’s temporary working memory for each turn, including things such as prompts, conversation history, tool results, and retrieved information.

Examples:

  1. The model can see parts of 20 project files at once.
  2. Conversation history is included in the Context.
  3. Web Search results are added to the Context.

AI Agent My-Journey-In-Codeless