FrontierSWE (Impl.)

coding agentscode

Modality: text · Language: en · Multilingual: no · Max score: 1

FrontierSWE (Impl.) evaluates software engineering implementation ability and reports model ranking on implementation tasks. Lower rank is better.

Methodology

Imported from llm-stats public benchmark metadata. Modality: text. Max score: 1. Categories: agents, code. Language: en.