Scale AI Blog

The Scale Research Team

SWE Atlas is Complete: Measuring Coding Agents Across the Engineering Loop

Can Coding Agents Become Engineers? We’re Finding Out.

MoReBench: Evaluating the Process of AI Moral Reasoning

The Remote Labor Index: Measuring the Automation of Work

SWE-Bench Pro: Raising the Bar for Agentic Coding

Testing & Evals

Testing & Evals

Advancing Agents: Introducing Scale’s Agentic Leaderboards

Actions, Not Words: MCP-Atlas Raises the Bar for Agentic Evaluation

Testing & Evals

Testing & Evals

TutorBench: Grading the Next Generation of AI Tutors

Using Rubrics to Build Better Models

The Future is Multilingual: Scale's New Evaluation Benchmark