Mellum2
Canonical version: Mellum2.
Mellum2 is JetBrains' open-weight LLM for AI workflows, released open source under Apache 2.0 in June 2026. It is the successor to Mellum: where the original focused on code completion, Mellum2 handles both natural language and code. JetBrains positions it as engineered for the hard parts of production AI: latency, throughput, and cost.
Architecture
- 12B total parameters, MoE design with only ~2.5B active per token
- Trained specifically on natural language and code; not multimodal
- Open weights on HuggingFace
What it's for
A small, cheap, fast model to do the unglamorous plumbing of agentic systems rather than frontier reasoning:
- Routing and orchestrating AI workloads
- Low-latency RAG pipelines
- Powering sub-agents in complex workflows
- Private, local AI deployment
References
- https://blog.jetbrains.com/ai/2026/06/mellum2-goes-open-source-a-fast-model-for-ai-workflows/
- https://huggingface.co/collections/JetBrains/mellum-2
Related
- Mellum
- JetBrains
- Junie (JetBrains)
- Large Language Models (LLMs)
- AI Open Weight Models
- AI Mixture of Experts (MoE)
- Apache 2.0 License
About Sébastien
Ready to get to the next level?
Found this valuable? Share it with someone who needs it.