Claude API
What is the maximum context window for each Claude model?
QUICK ANSWER
All current Claude 3 and Claude 3.5 models support a maximum context window of 200,000 tokens (approximately 150,000 words or 500 pages of text) per single API request.
Context Window Reference
| Model Name | Context Limit (Tokens) | Max Output Limit (Tokens) |
|---|---|---|
| Claude 3.5 Sonnet | 200,000 | 8,192 |
| Claude 3.5 Haiku | 200,000 | 8,192 |
| Claude 3 Opus | 200,000 | 4,096 |
Keep in mind that high token counts increase request latency. Implement prompt caching for files or systemic prompt libraries that span multiple requests.
Verified against: Anthropic Documentation 2026