Aeonic is an open-source AI/AGI/LLM router and orchestrator designed to intelligently route, compose, and manage calls across any combination of language models — local or cloud, small or frontier — with a unified API, zero vendor lock-in, and full observability.
Whether you're building a simple app that switches between cheap and powerful models based on task complexity, or orchestrating a galaxy-brain multi-agent pipeline across distributed statespaces, Aeonic gives you the primitives to do it cleanly.
| Capability | Description |
|---|---|
| Intelligent Routing | Semantically route prompts to the optimal model based on cost, capability, latency, and task type |
| Multi-Agent Orchestration | Compose agent pipelines with fan-out, fan-in, debate loops, and critic/verifier patterns |
| Distributed Statespaces | Manage context, vector stores, and episodic memory across distributed graph nodes |
| Provider Abstraction | Unified API across OpenAI, Anthropic, Gemini, Mistral, Ollama, Groq, Bedrock, and custom endpoints |
| Policy Engine | Declarative OPA/Rego-based routing policies — cost ceilings, compliance rules, rate limits as code |
| Observability | Full OpenTelemetry tracing, cost attribution, token accounting, and latency heatmaps |
┌─────────────────────────────────────────────────┐
│ Ingress Layer │
│ REST API · gRPC · WebSocket · Python · TS SDK │
└──────────────────────┬──────────────────────────┘
│
┌──────────────────────▼──────────────────────────┐
│ Core Orchestration │
│ Semantic Classifier → AEONIC ROUTER → State │
│ Policy Engine ↕ Cost Optimizer ↕ Fallbacks │
└──────────────────────┬──────────────────────────┘
│
┌──────────────────────▼──────────────────────────┐
│ Agent Layer │
│ Orchestrator → Workers → Memory/RAG → Critic │
└──────────────────────┬──────────────────────────┘
│
┌──────────────────────▼──────────────────────────┐
│ Provider Layer │
│ Anthropic · OpenAI · Gemini · Mistral · Ollama │
│ Groq · Bedrock · Custom Endpoints │
└─────────────────────────────────────────────────┘
-
Phase 0 — Core Router & Provider Abstraction
Unified API, provider adapters, basic cost-aware routing, OpenAI-compatible endpoint -
Phase 1 — Policy Engine & Semantic Routing (in progress)
OPA integration, task-type classifiers, declarative policies, latency-aware load balancing -
Phase 2 — Multi-Agent Orchestration (Q3 2025)
Native agent primitives, fan-out/fan-in, debate loops, shared statespaces -
Phase 3 — Distributed Memory & Long-Horizon Tasks (Q4 2025)
Vector-backed episodic memory, cross-session context, graph-of-thought execution -
Phase 4 — AGI Orchestration Primitives (2026)
Self-improving routing policies, recursive agent spawning, full statespace exploration
Note: Aeonic is in early development. The API is not yet stable. Star the repo to follow along.
# Clone the repo
git clone https://github.com/AeronicWarrior/aeonic.git
cd aeonic
# Install dependencies (coming soon)
pip install aeonicAeonic is built in the open and contributions of all kinds are welcome — code, docs, ideas, and bug reports.
- Fork the repo
- Create a feature branch:
git checkout -b feat/your-feature - Commit your changes:
git commit -m 'feat: add your feature' - Push and open a Pull Request
See CONTRIBUTING.md for full guidelines.
MIT — see LICENSE for details.