We ran Devin Fusion on our Code Migration benchmark and the performance of Devin Fusion with GPT-6 Astra and SWE-2 sidekick places it on the Pareto Frontier of the benchmark.
*Code Migration Bench is our benchmark that determines if models can reimplement working programs in another language.
benchmark — A standard public test set for comparing AI models — the shared scoreboards behind every "model X beats model Y" claim.
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
Why it matters
Evidence that pairing a cheaper sidekick model with a frontier model can match or beat the frontier model alone, at lower cost, is a concrete lever for reducing coding-AI agentAn AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.Full definition → spend without losing quality.
