SWE-Bench Multilingual

SWE-Bench Multilingual is a benchmark by the Princeton NLP group that evaluates code generation models on real-world software engineering tasks across multiple programming languages, with Cursor's Composer 2.5 achieving 79.8% on it at $0.50/M tokens.

Two-hop subgraph: this entity, every entity it directly relates to, and every entity those neighbors relate to. Drag a node, scroll to zoom, click to inspect — or click any neighbor and re-center the atlas there.