Launched this week
Jev Router
Smarter LLM routing. Lower inference costs.
1 follower
Smarter LLM routing. Lower inference costs.
1 follower
Jev Router is an open-source, self-hosted LLM routing layer that intelligently selects the right model for every request. Use one OpenAI-compatible API across multiple providers. Optimize for quality, cost, speed or privacy, with automatic fallbacks, built-in guardrails and full routing observability. Powered by Jev.
Free
Launch Team
Render Sandboxes: a computer for every agentIsolated compute, persistent state and a real filesystem
Promoted
Hey Product Hunt! 👋
I'm Sonu, founder of AI Anytime. We've been working on Jev Router, an open-source intelligent routing layer for LLM applications.
Here's the problem we wanted to solve:
Developers shouldn't have to manually choose between GPT, Claude, Gemini, DeepSeek or local models for every request.
Not every prompt needs the most expensive model. But sending a complex task to a lightweight model isn't the answer either.
We built Jev Router to make that decision intelligently.
What you get:
One OpenAI-compatible API for multiple LLM providers.
Automatic model selection using model="auto".
Five strategies: Balanced, Quality, Cheapest, Fastest and Privacy.
Automatic fallbacks when models fail.
Built-in guardrails for prompt injection, harmful content and PII.
Detailed routing traces, cost and latency analytics.
Self-hosted deployment with an MIT license.
Our initial evaluation:
Across 30 test prompts, Jev Router achieved 100% right-sized model selection and 35.6% cost savings compared with always using the strongest model.
We're releasing this openly because we believe intelligent model routing should be accessible to every developer.
Try it, inspect the code and run your own evaluations.
I'd particularly love feedback from developers building production-grade AI agents, RAG systems and multi-model applications.
What would you improve?