How it works
Claude Opus 4.7 introduces a new tokenizer (~1.35x token count for the same text), 1M context window, 128k max output, adaptive thinking only (no extended thinking budgets), and an xhigh effort level. API model IDs: claude-opus-4-7, claude-sonnet-4-6, claude-haiku-4-5.
Example
An agentic coding system uses claude-opus-4-7 for the planning and architecture decisions, claude-sonnet-4-6 for the bulk of code generation, and claude-haiku-4-5 for simple file operations and routing decisions.
