#142 · Primary category： LLM Application Frameworks
MoA
Together Mixture-Of-Agents (MoA) – 65.1% on AlpacaEval with OSS models
Project last updated：01/07/25
GitHub Stars
3.0K
Forks
386
Contributors
9
License
Apache-2.0
Why we included this project
Most LLM pipelines settle for whatever a single model answers, but MoA is a clean reference implementation of a different idea: send the question to several open-weight models, then let a separate aggregator model combine their responses into one stronger answer. Stacking that composition across layers squeezes out more quality, and the authors report 65.1% on AlpacaEval 2.0, ahead of GPT-4 Omni's 57.5%, using only open-source models. The repo keeps the pattern easy to borrow, with a ~50-line starter, a multi-layer example, and an interactive CLI chatbot, so you can watch the whole flow work before wiring it into your own code.
Articles for this project
No articles for this project yet.
To suggest a topic or contribute an article, contact us.
Related projects in this category
langchain
The agent engineering platform.
dify
Build Agentic workflows, RAG pipelines, with rich AI model and tool support on one collaborative workspace. Deploy on cloud, VPC, or self-hosted, so teams move from prototype to production without rebuilding the stack.
headroom
Compress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.
litellm
The fastest, litest AI Gateway. Rust core with Python SDK. Call 100+ LLM APIs in OpenAI (or native) format with cost tracking, guardrails, load balancing, and logging [Bedrock, Azure, OpenAI, Anthropic, OpenAI, VertexAI, vLLM, Nvidia NIM]
llama_index
LlamaIndex is the leading document agent and OCR platform