llm-ensemble 开源项目推荐（27 个）

algorithmicsuperintelligence /openevolve

AlphaEvolve 的开源实现

shyamsaktawat /OpenAlpha_Evolve

OpenAlpha_Evolve是一个开源Python框架，其灵感来自DeepMind的AlphaEvolve等自主编码代理的突破性研究。

baidu-baige /LoongFlow

LoongFlow is an expert-grade Agent framework for Loop Engineering. Through a Plan-Execute-Summary loop and structured experiential memory, it enables AI to continuously think, execute, reflect, and evolve across complex software engineering, mathematical, and machine learning tasks.

junchenzhi /Awesome-LLM-Ensemble

A curated list of Awesome-LLM-Ensemble papers for the survey "Harnessing Multiple Large Language Models: A Survey on LLM Ensemble"

inter-co /science-codeevolve

CodeEvolve is an open-source evolutionary coding agent for algorithm discovery and optimization.

trekhleb /yesbrainer

🧠 A council of AI models for the decisions that aren't no-brainers — they answer in parallel, debate to consensus, or get judged to a verdict. Browser-only, open source, bring your own keys (BYOK), no backend.

sreenathmmenon /llmswap

Ask, compare, and synthesize across 10 LLM providers. Python SDK, CLI, local Arena, MCP, and privacy-aware

0xAkuti /ai-council-mcp

Multi-AI consensus MCP server that queries multiple AI models (OpenAI, Claude, Gemini, custom APIs) in parallel and synthesizes responses to reduce bias and improve accuracy. A Python implementation of the wisdom-of-crowds approach for AI decision making.

magent4aci /openJiSi

[ICML2026] The official code of "Beyond Gemini-3-Pro: Revisiting LLM Routing and Aggregation at Scale"

luckeyfaraday /fusion-engine

Self-hosted Fusion API and OpenRouter Fusion alternative for building reliable multi-model LLM ensembles — fan out one prompt to N models, then synthesize one answer with a judge model. OpenAI-compatible API, CLI, and eval harness.

MeetAhalpara /NityaGeeta

Open-source Multi-LLM Ensemble & Scripture RAG platform delivering Bhagavad Gita guidance with zero-hallucination, page-level citations across 5,000+ commentary pages. Built with FastAPI, Next.js 15, and Weaviate.

albertofettucini /CouncilKit

Multi-model deliberation as a Swift primitive — ask several LLMs one question, blind peer review, a divergence score, and the dissenting view.

yoon6503 /SAFE

Official code for When to Ensemble: Identifying Token-Level Points for Stable and Fast LLM Ensembling (ICLR'26)

norbertm2050 /agent-huddle

Agent Huddle: multi-model AI discussion and LLM deliberation workspace for layered reasoning, AI agents workflows, and OpenAI-compatible model orchestration.

surajvs2710-stack /god-llm-decisions

Fuses two System-1 decision models (Laya + Jev) into one auditable verdict — confidence-weighted fusion, disagreement escalation to a local adjudicator, append-only JSON decision logs.

arthurthestupid /council-proxy

OpenAI-compatible multi-LLM ensemble proxy with Chairman fusion. Drop-in for any chat completions client.

vadimchernets /docprep

One AI assistance architecture, stated so it can be run: isolated generation, blind consolidation, and two critics from different vendors. Accompanies SSRN 7120940.

chirindaopensource /fused_llm_router

End-to-end Python research pipeline replicating "Beyond Benchmark Rankings": treats LLM selection as portfolio construction rather than leaderboard ranking. Estimates marginal utility, informational novelty, and correlated-failure risk per model, then runs budget-feasible greedy optimization to select LLM ensembles.

postman-cs /unifusion

A Claude Code skill that fans one hard question out to several frontier models (GPT-5.5, Gemini, Kimi, GLM) blind and in parallel, then has Opus 4.8 judge and synthesize one grounded answer. Multi-model test-time scaling with independence enforced and the judge kept separate from the panel.

lehighriverscalylentinus811 /LoongSage

Accelerate frontier LLM post-training with a production-grade, agentic reinforcement learning framework validated for scalable, high-efficiency training.