A new open-source AI leaderboard, whatmodel.sh, has been launched to help users determine which AI models can run on their local hardware. The tool takes user-provided RAM and GPU specifications to rank and suggest models, even comparing them against API-based models. Notably, a 32GB Mac can now run models like Qwen3.8 27B at 4-bit precision, achieving performance comparable to GPT-5, entirely locally and at no cost. The leaderboard's scoring methodology is based on over 19,000 benchmark results, with its data, scoring code, and benchmarking scripts available on GitHub. AI
IMPACT Enables users to identify and run high-performing AI models on their personal hardware, democratizing access to advanced AI capabilities.
RANK_REASON This is a new tool release, not a frontier model release or significant industry event.
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
