- Category
- AI Tools
- Rank
- No. 1384Tools index
Previous survey · No. 1319 ·
- Pricing
- Open Source
- Type
- TOOL
- Use case
- Models: Train & Run
- Interfaces
- API
- GitHub
- 83 stars
- Date
About
Gutsy is a local decision model that takes a state and typed questions (yes/no, choice, or score) and returns a probability for every option. It is a fine-tune of Qwen3.5-0.8B served as a 775 MB GGUF through llama.cpp on a CPU, with no text generation, network calls, or per-call cost. Its API follows the Jev-style Decisions format, so existing clients mostly need only a new base URL.
At a glance
Gutsy runs small local models that return scores for fixed decision options. It is worth benchmarking if you want repeated classifications on a CPU and can evaluate the scores on labeled examples.
Good fit
Developers experimenting with local triage or routing where options can be specified up front.
Consider something else when
Use another model if your task depends on multilingual input, broad knowledge or multi-step rulebooks.
Start here
Choose a small labeled set, shuffle the answer order and compare errors and confidence. Check how often the model should abstain before choosing a threshold.
Cost and requirements
Apache 2.0. The project documents a llama.cpp-based runtime and GGUF weights. It can run on a CPU with sufficient memory.
Limits to know
Calibration figures come from the project’s evaluation sets and can change with data or quantization. Verify the selected build on your workload.
Sources checked 2026-11-05. Source-based assessment.
Tags
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.