π€ AI providers can advertise thousands of models, but advertised doesn't mean working. π ModelScout discovers models, π§ͺ tests them with real requests, β€οΈ checks health, and β‘ measures latency. π It builds a local registry of models that actually work. DISCOVERED β WORKING π Open source β’ Lightweight β’ Provider-neutral π Discover. Test. Find what works.
Hey Product Hunt π
I built ModelScout because thereβs a simple problem with AI APIs:
An API can advertise thousands of models, but that doesnβt mean those models actually work.
π ModelScout discovers advertised models
π§ͺ Sends real lightweight probes
β€οΈ Checks their health
β‘ Measures latency
π Builds a registry of models that are actually working
The core idea is simple:
DISCOVERED β WORKING
ModelScout is open source, provider-neutral, and built around OpenAI-compatible APIs.
Iβd love to hear from developers building with multiple AI providers:
How do you currently figure out which models are actually usable?
π Discover. Test. Find what works.