Overview
The llm-anthropic plugin adds support for Claude's new Opus 4.7 model with thinking capabilities that let you see the AI's reasoning process. The update also improves token limits and removes obsolete headers for better performance.
Changes by Version
0.25
- Added claude-opus-4.7 model with thinking_effort support - enables extra-high reasoning capability
- Added thinking_display and thinking_adaptive options - lets you see AI reasoning in JSON output
- Changed max_tokens to maximum allowed per model - removes artificial token limits
- Fixed removed obsolete structured-outputs beta header - improves compatibility with older models