ResearchBenchmarksSoftware Engineering on SWE-bench 14 tasksFollow84.6Success RateS-BUS31.24845.09958.9572.801May 16, 2026Evaluation ResultsMethodMethodLinksSuccess RateCWRConflictsS-BUSN=4, Backbone=Llama-3....N=4, Backbone=Llama-3.1-8b-instant (Groq)2026.0584.60.190S-BUSN=8, Backbone=Llama-3....N=8, Backbone=Llama-3.1-8b-instant (Groq)2026.0572.20.180LangGraphN=8, Backbone=Llama-3....N=8, Backbone=Llama-3.1-8b-instant (Groq)2026.0569.49.340LangGraphN=4, Backbone=Llama-3....N=4, Backbone=Llama-3.1-8b-instant (Groq)2026.0533.37.610