Show HN: Find the best local LLM for your hardware, ranked by benchmarks

Selecting the right local LLM for your hardware setup has traditionally required extensive trial-and-error benchmarking. A new open-source project aims to eliminate this friction by providing a ranked comparison of local models based on actual performance metrics across different hardware configurations.

This tool is particularly valuable for practitioners deploying on constrained hardware—whether that's consumer GPUs, edge devices, or CPU-only systems. By aggregating benchmark data and making it queryable by hardware specs, the project reduces the time to find production-ready models and helps developers avoid costly mistakes in their deployment decisions.

The benchmarking approach matters because local LLM performance varies dramatically based on quantisation level, batch size, context length, and hardware capabilities. Having this centralized reference means the community can make data-driven choices rather than relying on anecdotal reports or vendor claims.

Source: Hacker News · Relevance: 9/10