🚀 Open Source · MIT License · 556 Tests Passing

Never Let Your
LLM Calls Fail

Generate, test, and optimize intelligent fallback chains for any LLM provider. One config, instant resilience. Free forever.

View on GitHub → See How It Works
556
Tests
15
API Endpoints
5+
Export Formats
0
Dependencies on LLM SDKs
chain.yaml
# Define your fallback chain in seconds name: production-chain providers: - name: GPT-4 model_id: gpt-4 priority: 0 - name: Claude model_id: claude-3-sonnet priority: 1 fallback_rules: - condition: rate_limit action: retry max_retries: 2 - condition: timeout action: failover

Built for Reliability

Everything you need to make your LLM integrations bulletproof — all in one tool.

🔗
Smart Fallback Chains
Define priority-ordered provider chains with automatic failover. When one provider fails, the next takes over instantly.
🧪
Simulated Testing
Test your chains against outages, rate limits, and timeouts before they happen in production. No real API calls needed.
📦
Multi-Format Export
Export to LiteLLM, OpenRouter, LangChain, Haystack, or custom Jinja2 templates. Drop-in integration with your stack.
⚙️
Rule-Based Routing
Configure retry, failover, wait, and custom actions per error type. Rate limits get retried, timeouts failover, errors escalate.
📊
Live Dashboard
Dark-themed web UI for creating chains, running tests in real-time via WebSocket, and exporting configs. Zero config.
🔑
Auth & Rate Limiting
Built-in API key authentication, per-IP rate limiting, and token bucket controls. Production-ready security out of the box.

How It Works

Three steps to bulletproof LLM reliability.

1

Define Your Chain

Create a YAML config with your LLM providers in priority order and fallback rules for each error type.

2

Test & Simulate

Run simulated outages to verify your chain handles failures correctly. See latency, retries, and failover paths in real-time.

3

Export & Deploy

Export to your framework of choice. Drop the config into your app and let FallbackRabbit handle the rest.

Works With Your Stack

Supports all major LLM providers and frameworks.

🟢 OpenAI
🟠 Anthropic
🔵 Azure OpenAI
🦙 Ollama
🔧 Custom Endpoints
📦 LiteLLM
🦜 LangChain
🌾 Haystack
🌐 OpenRouter
📄 Custom Jinja2

100% Open Source

Free forever. MIT licensed. No usage limits, no paywalls, no telemetry.

📜
MIT License
Use it anywhere — commercial, personal, whatever.
🧪
556 Tests
Comprehensive test coverage across all modules.
🐍
Python 3.11+
Tested on 3.11 and 3.12. Type-hinted throughout.
🐳
Docker Ready
One command: docker compose up
🤝
Contributions Welcome
See CONTRIBUTING.md. All skill levels welcome.
Contribute →
📚
Full Docs
MkDocs Material documentation with examples.
Read Docs →

Ready to Make Your LLM Calls Bulletproof?

Get started in under 5 minutes. No credit card, no signup, no limits.

View on GitHub → Read the Docs