According to Greg Brockman (@gdb), a new benchmark has been released to evaluate the capability of AI systems in expert-level scientific reasoning, signaling a major leap in scientific progress through AI in 2026. This benchmark provides standardized metrics to assess how well AI models can perform complex scientific tasks, helping organizations gauge AI readiness for research applications and accelerating innovation in scientific fields. The introduction of such a benchmark is expected to drive investment in AI-powered research tools and enable businesses to identify opportunities in AI-driven scientific discovery (source: Greg Brockman via Twitter, Dec 16, 2025).