Claude Opus 4

Claude Opus 4 is a model in Anthropic's Claude model family, referenced in the context of Claude Code development and self-evaluation workflows.

Use in Claude Code Self-Testing

In a reflection on a year of Claude Code development, it was noted that Claude Opus 4 was used to test itself — the team "hooked it up to Opus 4" so the model could run its own evaluations. Reflecting on a year of Claude Code [2:30] This suggests Opus 4 was integrated into automated evals or self-assessment pipelines during Claude Code's development cycle, representing an approach where a capable frontier model is used to evaluate its own outputs or behaviors.