Show HN: Gutsy – Tiny model for typed decisions on CPU (Jev style)

Category: ai-ml

Tags: local-inference, calibrated-probabilities, decision-model, cpu, quantized-model, structured-output

Score: 7.5/10 (Innovation: 7, Technical: 8, Documentation: 8, Utility: 7)

Gutsy is a tiny 0.8B-parameter language model fine-tuned for typed decisions (yes/no, choice, score) that runs locally on CPU via llama.cpp. It outputs calibrated probabilities, follows the Jev-style API, and competes with larger models on decision benchmarks while being private and deterministic.

Target audience: backend devs, ai engineers, ml engineers, privacy-conscious teams

Repository: https://github.com/kouhxp/gutsy · HTML · Apache-2.0 · 94 stars

View on Hacker News