gutsy
Local CPU decision model that returns calibrated probabilities for yes/no, choice, or score questions.
Tested on Linux
Use it
Linux x86_64
These steps ran from a folder named work in the home folder, which already existed. Make it and run them from there:
mkdir -p "$HOME/work" && cd "$HOME/work"git clone the gutsy repo (source for gutsy-inference)
cd "$HOME"/work && git clone --depth 1 https://github.com/kouhxp/gutsycreate venv in HOME and pip install -e gutsy-inference (builds llama-cpp-python 0.3.37 from source)
python3 -m venv ~/gutsy-venv && ~/gutsy-venv/bin/pip install -e "$HOME"/work/gutsy/gutsy-inference
Verified in pickfit: the commands exactly as its sandbox ran them
For the record only: they name the sandbox's home, /home/sbx.
cd /home/sbx/work && git clone --depth 1 https://github.com/kouhxp/gutsypython3 -m venv ~/gutsy-venv && ~/gutsy-venv/bin/pip install -e /home/sbx/work/gutsy/gutsy-inference~/gutsy-venv/bin/gutsy-inference --helpCheck that it works
~/gutsy-venv/bin/gutsy-inference --helpThe output should contain “usage: gutsy-inference”.
What it does
- Get calibrated probabilities for every optionNot tested
- Ask yes/no, choice and score questionsNot tested
- Choose among up to 255 options with rejectNot tested
- Ask cheap follow-up questions on one stateNot tested
How far pickfit got
Linux x86_64
- Installs
- Starts
- 0 of 4 features work
- Works offline
Good for
- Triage a support ticket for team, refund and mood
- Ask yes/no questions of feedback rows
- Pick the cheapest model per message
- Score each claim against a source
- Detect live triggers on a sales call
An alternative to Laya and Jev.
How pickfit tested it
Each run starts from a clean machine and follows the steps above. How pickfit verifies
Linux x86_64
Installs and starts
- Replayed from scratch
- Works offline
- Checked on 16 Nov 2026
Differences the test machine hit
- README: pip install -e gutsy/gutsy-inference # needs llama-cpp-python >= 0.3.35
In the test: Installed into a venv under $HOME (python3 -m venv ~/gutsy-venv, then ~/gutsy-venv/bin/pip install -e). The plain pip command was not run here, so no claim is made about PEP 668 on this system Python. llama-cpp-python had no prebuilt wheel for this box and was compiled from source (about 3.5 minute… - README: hf download kouhxp/gutsy --include "gutsy-0.8b-v04*" --local-dir gutsy/gutsy-inference/models
In the test: Not run. The model files are 505 MB (Q4_K_M) and 775 MB (Q8_0), above this box's ~300 MB download cap, so gutsy-inference check, serve and all README API examples were not run.