Intelligent routing for your agents

Your agentic workloads don't need one expensive LLM but many cheaper LLMs

How it works

1

Connect

Use Switchyard as a model provider. OpenAI-compatible endpoint that works with any SDK.

2

Route

Our intelligent router seamlessly switches to a cheaper model based on your usage and requirements. Simple queries go to fast/cheap models, complex ones to the default model.

3

Done

That's it. No changes needed in your stack. Our router works silently under the hood with <10ms latency.

Why Switchyard

Stop overpaying for simple queries

Not every prompt needs a large expensive model. Our classifier routes simple queries to cheaper models automatically.

One key, every provider

Bring your own keys and route across providers intelligently. No need to manage multiple API keys.

Easy to use

OpenAI-compatible API, works with any SDK, no complicated setup needed. Drop it in and start saving.

Pricing

Self-serve (BYOK)

Free

Install on your machine. Bring your own API keys, get smart routing.

  • Full routing capabilities
  • Local inference
  • No usage limits
Get started

Cloud

Pay-as-you-go

Load funds, use inference. No monthly minimums.

  • Managed inference
  • No API keys needed
  • Pay only for what you use
Join waitlist