Context window
Published Updated
What is Context window?
A context window is the maximum amount of text, measured in tokens, that an AI model can consider at once. Anything beyond it is cut off or forgotten, which is why long agent sessions lose track of early instructions.
The maximum amount of text (measured in tokens) an AI model can process and remember in a single interaction.
A context window dictates how much information you can feed an AI model at one time. If a conversation or document exceeds this limit, the model will "forget" the earliest information provided. Modern frontier models support context windows of 1 to 2 million tokens, enabling the ingestion of entire codebases.
How it relates to Orbit
Orbit’s Agent Sessions show context consumption from the live stream, and Apollo offers a handoff when its conversation fills up. See Complexity Tiers & Routing for details.
When you would use it
You constantly monitor the context window to ensure your agent doesn't "forget" its original instructions during a long, multi-step orchestration task.
Common operations
- Pasting entire software repositories into a prompt for architectural review.
- Maintaining long-running conversational agent loops without losing state.
Frequently asked questions
How big is a context window?
It depends on the model. Current frontier models handle hundreds of thousands to a few million tokens, which can be enough for a whole codebase.
What happens when the context window fills up?
The earliest material falls out of view or is summarized, so an agent can forget its original instructions mid-task.
How does Orbit handle a full context window?
Agent Sessions show context consumption live, and Orbit carries each project’s context between sessions so a fresh session starts informed. See Context & Handoffs.