Models · MarkTechPost ·
JetBrains Releases Mellum2: A 12B MoE Model for Fast, Specialized Tasks in Multi-Model AI Pipelines
JetBrains has released Mellum2, a 12B Mixture-of-Experts model trained on 10.6 trillion tokens for fast, specialized tasks in multi-model AI pipelines. The model is open-sourced under the Apache 2.0 license.