What Is a Context Window?
The AI "memory limit" that trips people up — explained simply
A context window is the maximum amount of text an AI model can consider at once, measured in tokens. Think of it as the model's short-term memory: everything in your prompt, plus the conversation history and the model's reply, all has to fit inside it. Go over the limit and the oldest text gets dropped — the model simply "forgets" it.
Context windows of popular models
| Model | Context window | Roughly equals |
|---|---|---|
| GPT-3.5 Turbo | 16K tokens | ~12,000 words |
| GPT-4o | 128K tokens | ~96,000 words |
| Claude 3.5 Sonnet | 200K tokens | ~150,000 words |
| Gemini 1.5 Pro | 1M tokens | ~750,000 words |
Rule of thumb: 1 token ≈ ¾ of an English word, or ~4 characters.
Why it matters
- Truncation: If your document exceeds the window, part of it is ignored — and the model won't tell you which part.
- Cost: You pay per token sent. A bigger window lets you send more, but every token still costs money.
- Quality: Even within the limit, models can lose track of details buried in the middle of a very long prompt (the "lost in the middle" effect).
How to stay within the limit
- Trim boilerplate and repeated instructions from your prompt.
- Summarize long documents before sending them.
- For chats, drop or summarize old turns you no longer need.
- Check your token count before you send — our token counter shows exactly what percentage of a model's context window your text uses.