Show HN: Gutsy, a 0.8B Jev-compatible decision model that runs on your CPU

Category: ai-ml

Tags: local-inference, decision-making, calibration, cpu-inference, llama-cpp, qwen-finetune

Score: 7.0/10 (Innovation: 7, Technical: 7, Documentation: 8, Utility: 6)

Gutsy is a 0.8B-parameter fine-tune of Qwen3.5-0.8B that provides calibrated probability outputs for yes/no, choice, and score decisions, running locally on CPU via llama.cpp. It offers a Jev-compatible API, deterministic inference, and strong calibration, making it a practical drop-in for local decision-making without network calls or per-call costs.

Target audience: backend developers, AI engineers, data scientists, and teams needing private, cost-effective decision APIs

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

View on Hacker News