Instructions to use fly88oj/decidex-gguf with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use fly88oj/decidex-gguf with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf fly88oj/decidex-gguf:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf fly88oj/decidex-gguf:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf fly88oj/decidex-gguf:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf fly88oj/decidex-gguf:Q4_K_M

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf fly88oj/decidex-gguf:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf fly88oj/decidex-gguf:Q4_K_M

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf fly88oj/decidex-gguf:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf fly88oj/decidex-gguf:Q4_K_M

Use Docker

docker model run hf.co/fly88oj/decidex-gguf:Q4_K_M

LM Studio
Jan
Ollama
How to use fly88oj/decidex-gguf with Ollama:
```
ollama run hf.co/fly88oj/decidex-gguf:Q4_K_M
```
Unsloth Desktop

How to use fly88oj/decidex-gguf with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf fly88oj/decidex-gguf:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "fly88oj/decidex-gguf:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use fly88oj/decidex-gguf with Docker Model Runner:
```
docker model run hf.co/fly88oj/decidex-gguf:Q4_K_M
```

Lemonade

How to use fly88oj/decidex-gguf with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull fly88oj/decidex-gguf:Q4_K_M

Run and chat with the model

lemonade run user.decidex-gguf-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use fly88oj/decidex-gguf with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf fly88oj/decidex-gguf:Q4_K_M

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default fly88oj/decidex-gguf:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use fly88oj/decidex-gguf with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf fly88oj/decidex-gguf:Q4_K_M

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "fly88oj/decidex-gguf:Q4_K_M" \
  --custom-provider-id llama-cpp \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

Decidex GGUF builds — decision engine for llama.cpp / Ollama / LM Studio

Quantized builds of the decidex-core-8b adapter merged into Qwen3-8B:

Quantized builds carry a lineage marker (r1 = first release of the core-8b adapter line, internal training round 7).

File	Size	Best for
`decidex-core-8b-r1-Q4_K_M.gguf`	5.0 GB	16GB GPUs / 16GB+ RAM CPU
`decidex-core-8b-r1-Q6_K.gguf`	6.7 GB	quality-first
`decidex-core-8b-r1-Q8_0.gguf`	8.7 GB	closest to BF16 behavior

Companion MTP draft: decidex-draft-0.6b.

How to use (decision engine, not chat)

Send the completion prompt below, temperature 0, 1–4 max tokens, and read the first letter — or read letter logprobs for the full probability distribution (llama-server n_probs). Do NOT use a chat template (the base model is a thinking hybrid; chat templates break the readout).

You are a precise decision engine. Evaluate the STATE against the
QUESTION and pick the single best option.

STATE:
{your state}

QUESTION:
{your question}

OPTIONS:
A. {option A}
B. {option B}
C. {option C}

Answer with the letter of the single best option.
Answer:

Verified live through a CPU llama-server: official-docs example returns noul 0.9503 (official 0.95) and score 1.0404 (official 1.05). Ready-made client: examples/gguf_decision_client.py in the Decidex repo; full guide: GGUF-USAGE.md there. Not affiliated with TypeSafe AI.

Downloads last month: 105

GGUF

Model size

8B params

Architecture

qwen3

Hardware compatibility

4-bit

6-bit

8-bit

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for fly88oj/decidex-gguf

Base model

Qwen/Qwen3-8B-Base

Finetuned

Qwen/Qwen3-8B

Quantized

(451)

this model