Beholder — GGUF (native / local)
GGUF build of the Beholder state-extractor for native local runtimes — llama.cpp, LM Studio,
KoboldCpp, Ollama, and any OpenAI-compatible local server.
Beholder reads roleplay / narrative prose and emits structured character state — clothing, colors,
materials, held items, and wounds — per body slot.
Files
Beholder-Q8_0.gguf — 8-bit quant (recommended; best quality/size for a model this small)
Running in the browser instead? Use the WebGPU build →
GetBeholder/Beholder-q4f16.
License
PolyForm Noncommercial 1.0.0. Commercial use by permission.