Scout: find a model for what you’re building
Scout finds model candidates on Hugging Face for a task you describe in a sentence. It searches live repository records, reads model cards, and compares task fit, language metadata, licenses, estimated weight memory and available inference providers. Each recommendation links to its evidence, with a setup snippet and a shareable search.
How it works
- Describe the job: the task, the language, where it runs and how much memory you have. Refine the search pins the task, language, memory budget, license or deployment when Scout guesses wrong.
- Scout searches the Hub, reads up to 12 model cards and checks each candidate against your purpose, language, deployment preference, memory budget and license filter.
- Each pick comes with a quote from its model card, the checks it passed, its estimated FP16 weight memory, inference providers, open questions and a code snippet to try it.
- To run a chat model on your own machine, describe the computer: Scout says which GGUF build fits it, whether Ollama, LM Studio and llama.cpp run on it, the commands to copy, and an Ollama Modelfile with your own system prompt and settings.
Privacy and limits
- Your request is processed on the server; Scout only reads public model records.
- Memory is screened by the estimated FP16 weight size. Running a model needs more, and quantized versions need less.
- Scout hasn’t run or benchmarked these models: try them on your own examples.
More
Scout runs on this site’s desktop, which needs JavaScript: with it on, this address opens the app. In the desktop’s terminal it’s app scout. This page as Markdown.