Local LLM Model Finder

Instantly discover which large-language models you can run on your PC. Filter by GPU, VRAM, RAM, CPU & OS — explore specs, see a 3D scatter of release age vs. size vs. tokens, and browse the full table with universal filter.

We applied our 'Optimal Rating' for assesing specs and characteristics of local LLMs:

  • Input: 1 to 2 points according to the input_scores dictionary (maximum 2 points for the combined "Text, Image" type).
  • Token Count (Token_Count_Thous) (0 to 5 points by normalizing the current value relative to the minimum and maximum in the dataset) (if all models have the same number of tokens, 2 points are automatically assigned).
  • Size (Size_GB) (0 to 4 points by inverse proportion (the smaller the model, the higher the score, with 2 points automatically assigned for equal sizes).

    • After which the sum is rounded to one decimal place and limited to the range from 0 to 10 points, which provides a final score from 0 (worst performance) to 10 (best performance) taking into account all three evaluation criteria.

    All Models