Composer 2 Technical Report

Benchmark Model Rank Results
agents-on-frontiersweComposer 2.5 (Cursor CLI)–Dominance: 40.27Average Rank: 8.76
code-generation-on-swe-bench-multilingualComposer 2 (Cursor harness)–Accuracy: 73.7
code-generation-on-swe-bench-multilingualComposer 2.5 (Cursor harness)–Accuracy: 79.8