LiveSWEBench (https://liveswebench.ai/) is a benchmark designed to evaluate the software engineering capabilities of AI agent applications.

This question ask about top average score in "Agentic Programming" AND "Target Editing" combined. Top score at 1 April 2025 is 47.83 (SWE-Agent with Claude Sonnet 3.7).

Will be judged according to the official leaderboard.