Coding & software engineering

A model's score is built only from results that someone other than its maker ran.

• The scoring set is the benchmarks chosen for this specialty. A benchmark takes part only when at least 5 models have an independent or third-party result on it.
• Each result is rescaled to 0 to 1, and a model's best setting counts.
• The score is the average over the benchmarks a model has, times 100.
• A model ranks only when it has results on enough of them. The rest show "Not enough independent results yet".
• Self-reported results are shown but never counted.

Method: category-2026-09-29.1

Scoring set: CursorBench, DeepSWE, FrontierCode, FrontierSWE, SWE-rebench, Terminal-Bench, MirrorCode, Aider Polyglot, GSO.

Effective set (used for the score today): DeepSWE, FrontierCode, MirrorCode, Aider Polyglot, GSO

• CursorBench: no importable independent results yet
• FrontierSWE: no importable independent results yet
• SWE-rebench: no importable independent results yet
• Terminal-Bench: no importable independent results yet

• Aider PolyglotActiveNear-ceilingCounts toward the score67 results
• CadEvalUnmaintainedHeadroom14 results
• CursorBenchActiveHeadroomCounts toward the score0 results
• CybenchActiveSaturated21 results
• DeepSWEActiveHeadroomCounts toward the score69 results
• ExploitBenchActiveHeadroom18 results
• FrontierCodeActiveHeadroomCounts toward the score34 results
• FrontierSWEActiveHeadroomCounts toward the score0 results
• GSOActiveHeadroomCounts toward the score37 results
• METR Time HorizonsActiveNear-ceiling0 results
• MirrorCodeActiveHeadroomCounts toward the score9 results
• PostTrainBenchActiveHeadroom12 results
• SWE-bench ProActiveSaturated0 results
• SWE-bench VerifiedSupersededSaturated34 results
• SWE-rebenchActiveHeadroomCounts toward the score0 results
• Terminal-BenchActiveHeadroomCounts toward the score0 results
• Terminal-Bench 2.0SupersededNear-ceiling198 results
• WeirdML (v2)ActiveSaturated166 results