Report Abuse

Basic Information

MassGen is an open-source multi-agent orchestration system for GenAI that coordinates multiple large language model agents to solve complex tasks through parallel, iterative collaboration. It provides an orchestrator, a shared collaboration hub, and a notification system so diverse agents can run in parallel, share intermediate work, learn from each other, and converge on higher-quality results than single models. The repository includes a CLI, YAML/JSON configuration format for single or multi-agent teams, interactive multi-turn mode, real-time terminal displays, comprehensive logging, and support for both API-based providers and local inference via LM Studio. It is intended for researchers and developers who want to experiment with or deploy multi-model teams using providers such as Anthropic Claude, Google Gemini, xAI Grok, OpenAI GPT series, Cerebras, Z AI and various open-weight models, and to extend or instrument agent collaboration patterns.

Links

Categorization

App Details

Features
MassGen emphasizes cross-model agent synergy, parallel processing, intelligence sharing, consensus building, and live visualization. It supports many API providers and local models, including Cerebras GPT-OSS-120B, Claude Sonnet and Haiku, Gemini variants, Grok, OpenAI GPT-5 family, Z AI GLM-4.5, and LM Studio local models. Built-in tool support varies by backend and can include web search, code execution, file operations, and native Claude Code tools like Read Write Edit Bash and notebook editing. System features include convergence detection and adaptive coordination so agents can restart or refine based on peers, multiple UI display types, comprehensive session logging, quick-start CLI commands for single-model or multi-agent runs, and a modular configuration format for detailed backend and agent settings.
Use Cases
MassGen helps teams and researchers obtain more robust and well-rounded outputs by enabling multiple models to work together rather than relying on a single model. It is useful for complex question answering, research tasks, creative writing, development and coding workflows that can leverage Claude Code and other coding tools, and scenarios that benefit from model diversity and consensus. Real-time displays and detailed logs improve observability and reproducibility of agent sessions. Local model support via LM Studio reduces inference cost and enables hybrid local/API setups. The configurable YAML/JSON agent definitions and CLI make it straightforward to prototype, compare approaches, and extend the system with new models, tools, and visualization or web UI components.

Please fill the required fields*