Mellum2

Trends

• Open-weight models diversify in architecture, modality, and size

Items

1. Jun 1

   JetBrains' Mellum2

   A 12B Mixture-of-Experts model (2.5B active) trained from scratch on natural language and code under Apache 2.0, built for latency-sensitive routing, RAG, and sub-agent work with more than 2x faster inference than similar-sized models.