hi

Other models, same or similar prompt

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.