Claude API
What is the maximum output token limit for Claude models?
QUICK ANSWER
Claude 3.5 Sonnet and Claude 3.5 Haiku support a maximum completion output of 8,192 tokens per single API request. Claude 3 Opus yields up to 4,096 output tokens.
Output Limits Overview
Ensure you configure the max_tokens parameter to match these thresholds if your application generates large documents, codebases, or analytical outputs. If a completion is truncated due to output limits, the API returns a response finish reason of max_tokens.
Verified against: Anthropic API Specifications 2026