Beta version: *Information might not be fully accurate. Please report any discrepancies.

SWE-bench Pro

swe-bench-proAgents & ToolsHigher is better

Higher-difficulty SWE-bench subset for frontier coding agents.

Loading chart…

Loading leaderboard…

Benchmark Info

Current SOTA

Best Open Source

Median Score52.3%

DistributionClustered

Score Distribution

P10

P50

P90

0100%

AgentBench

agentbench

IFEval

ifeval

Inverse IFEval

ifeval-inverse

IFBench

ifbench

Verified AdvancedIF

verified-advancedif

MultiChallenge

multichallenge