JetBrains' Mellum2.1 lifts SWE-bench score from 2.0 to 47.0 with pure RL

JetBrains released Mellum2.1, a 12B mixture-of-experts coding model that activates just 2.5B parameters per token. Reinforcement learning in real repositories lifted its SWE-bench Verified score from 2.0 to 47.0.