Context Window
The context window is the maximum amount of text — measured in tokens — that a language model can process at one time, including both the input prompt and the generated output.
Definition¶
The context window (also called the context length) is the maximum number of tokens a language model can attend to simultaneously. This includes the input (the user's prompt, any system instructions, and conversation history) and the output (the model's generated response). When input plus output exceeds the context window, the model can no longer "see" the earlier parts of the conversation.
Why Context Window Size Matters¶
A larger context window means: - Longer documents can be processed without splitting - Longer conversations can be maintained without the model "forgetting" earlier exchanges - More complex tasks with extensive instructions or examples can be handled in a single call
For practical writing tasks — humanizing a blog post, rewriting a section, checking grammar — context windows of current models are more than sufficient. The constraint becomes relevant when processing book-length documents or maintaining long multi-turn conversations.
Current Context Window Sizes (2025–2026)¶
| Model | Context Window |
|---|---|
| GPT-4o | 128,000 tokens (~96,000 words) |
| Claude 3.5 Sonnet | 200,000 tokens (~150,000 words) |
| Gemini 1.5 Pro | 1,000,000 tokens (~750,000 words) |
| Llama 3.1 (70B) | 128,000 tokens |
Context vs. Training Data¶
The context window is different from the model's training data. Training data is what the model learned from (a fixed corpus); the context window is what the model can actively reference during a specific interaction. A model with a 128K token context window still has knowledge from its training — the context window determines how much of the current conversation it can consider.