Show HN: CPU-first model for fast typed decisions

Category: ai-ml

Tags: llm, local-inference, calibration, cpu, decisions, llama-cpp

Score: 7.3/10 (Innovation: 7, Technical: 7, Documentation: 8, Utility: 7)

Gutsy is a CPU-first inference runtime and fine-tuned 0.8B model (Qwen3.5-based) that answers typed decision questions (yes/no, choice, score) with calibrated probabilities, avoiding text generation and network calls. It offers a local, deterministic, and privacy-preserving alternative to closed APIs for classification, routing, and judging tasks, with strong documentation including benchmarks, limitations, and browser demos.

Target audience: backend developers, ML engineers, data scientists, and product teams needing local, privacy-preserving decision models

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

View on Hacker News