PaperBench: Evaluating AI’s Ability to Replicate AI Research
We introduce PaperBench, a benchmark evaluating the ability of AI agents to replicate state-of-the-art AI research.
2 April 2025 · Source: OpenAI news
The full text is available at the source.