Background

Both models were released on the same day. Both are their respective companies' frontier reasoning models. This is as close to a fair fight as we could possibly hope for. So, who is ahead? OpenAI or Anthropic?

Resolution Criteria

Resolves according to the score shown for each model on https://metr.org/blog/2025-03-19-measuring-ai-ability-to-complete-long-tasks/ (50% time horizon)