Shared benchmarks

3
Shared benchmarks for Composer 1 and Composer 1.5
BenchmarkComposer 1Composer 1.5
  • 38.0%
  • 44.2%
Evaluation details for CursorBench 3.0

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Composer 1

Record for Composer 1 on CursorBench 3.0

Composer 1.5

Record for Composer 1.5 on CursorBench 3.0
  • 56.9%
  • 65.9%
Evaluation details for SWE-bench Multilingual

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Composer 1

Record for Composer 1 on SWE-bench Multilingual

Composer 1.5

Record for Composer 1.5 on SWE-bench Multilingual
  • 40.0%
  • 47.9%
Evaluation details for Terminal-Bench 2.0

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Composer 1

Record for Composer 1 on Terminal-Bench 2.0

Composer 1.5

Record for Composer 1.5 on Terminal-Bench 2.0