Anthropic Claude: Claude Opus 5 API
Claude Opus 5 is the current flagship of Anthropic's Opus line, built for highly autonomous agents, long-horizon engineering tasks and knowledge work. Adaptive thinking is on by default and reasoning depth is tuned through five `output_config.effort` levels, with a 1M-token context window and 128K single-response output. It also offers a fast mode that raises output throughput to roughly 2.5x at the same intelligence level.
- Context window: 1,000,000 tokens
- Max output: 128,000 tokens
- Input: text, image, file
- Output: text
- Reasoning: Supported
- Tool calling: Supported
- File input: Supported
- Released: 2026-07-24
- Knowledge cutoff: 2026-05-31
Frequently Asked Questions
What is the context window of Claude Opus 5?
Claude Opus 5 supports a context window of up to 1,000,000 tokens.
Does Claude Opus 5 support function calling?
Yes. Claude Opus 5 supports tool / function calling.
Does Claude Opus 5 support reasoning?
Yes. Claude Opus 5 is a reasoning-capable model.
What is the knowledge cutoff of Claude Opus 5?
The knowledge cutoff of Claude Opus 5 is 2026-05-31.