Gutsy is listed as Open Source. Check https://github.com/kouhxp/gutsy for the current plan limits, as vendors change pricing frequently.

Gutsy

Q: What is Gutsy?

Gutsy is a code AI tool listed on Use AI Tools. Gutsy runs local decision models that return calibrated probabilities. You send a state, text or JSON, and typed questions — yes or no, choice, or score — and get a probability for every option back, with no text generation, no network calls and no per-call cost. The API follows the Jev-style Decisions format so existing clients mostly need a new base URL. The model is gutsy-0.8b, a fine-tune of Qwen3.5-0.8B in Q8_0 or Q4_K_M, served by llama.cpp on an ordinary CPU. The state is encoded once, so each extra question costs only its own tokens, and there is a reject probability for none of these fits.

Q: What is Gutsy best for?

Gutsy is commonly used for Calibrated decisions with no API bill, Asking many questions about one state, Running decisions on CPU only. It supports English.

Code⚡ Advanced✅ Global access

Calibrated decisions with no API billAsking many questions about one stateRunning decisions on CPU only

Open SourceEnglish

Gutsy runs local decision models that return calibrated probabilities. You send a state, text or JSON, and typed questions — yes or no, choice, or score — and get a probability for every option back, with no text generation, no network calls and no per-call cost. The API follows the Jev-style Decisions format so existing clients mostly need a new base URL. The model is gutsy-0.8b, a fine-tune of Qwen3.5-0.8B in Q8_0 or Q4_K_M, served by llama.cpp on an ordinary CPU. The state is encoded once, so each extra question costs only its own tokens, and there is a reject probability for none of these fits.

Visit Website Compare

Community Rating

Be the first to rate Gutsy. Your feedback helps other users decide.

🎯 Real-World Use Cases

Serve it locally

Install the inference package, download the model and run the local HTTP service.

Ask typed questions

Send state plus yes/no, choice or score questions and read the probabilities per option.

Ask follow-ups cheaply

Because the state is encoded once, each additional question costs only its own tokens.

⚖️ Honest Review

✅ Pros

+Calibrated probabilities rather than generated prose
+No network calls and no per-call cost
+Deterministic on the same machine and build, and robust to option order

⚠️ Cons

−Requires downloading and running model weights locally
−The quoted calibration and order-robustness figures are the project's own

📚

Guides and comparisons about code tools

CodeOct 26, 2026