Instructions to use cosmospan/peek with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use cosmospan/peek 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 cosmospan/peek:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf cosmospan/peek:Q4_K_M

Install from WinGet (Windows)

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

Build from source code

git clone https://github.com/ggml-org/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-app
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama serve -hf cosmospan/peek:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama cli -hf cosmospan/peek:Q4_K_M

Use Docker

docker model run hf.co/cosmospan/peek:Q4_K_M

Llama new
LM Studio
Jan
Ollama
How to use cosmospan/peek with Ollama:
```
ollama run hf.co/cosmospan/peek:Q4_K_M
```
Unsloth Desktop

How to use cosmospan/peek with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf cosmospan/peek: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": "cosmospan/peek:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use cosmospan/peek with Docker Model Runner:
```
docker model run hf.co/cosmospan/peek:Q4_K_M
```

Lemonade

How to use cosmospan/peek with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull cosmospan/peek:Q4_K_M

Run and chat with the model

lemonade run user.peek-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use cosmospan/peek with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf cosmospan/peek: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 cosmospan/peek:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use cosmospan/peek with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf cosmospan/peek: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 "cosmospan/peek: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"

Peek: Squad Code's job router

Peek reads the job a person types into Squad Code and picks the kind of job (ask, build, fix, review, test, tidy) and the bot that should take it, with how sure it is and a short reason. It's Qwen3.5-0.8B fine-tuned with LoRA, then quantized to one 4-bit GGUF file (about 0.5 GB).

Trained on 1,859 examples from hand-written seeds plus 3,600 examples written and labelled by a teacher model.
Fixed test sets: 53/53 on route.jsonl, 120/120 on route_hard.jsonl. The hand-written rules score 51/53 and 70/120.
Runs on a laptop: about 250 MB of memory and about 0.4 s an answer on an M1.

Squad Code downloads the file named in latest.json when you turn the model on, and offers newer versions as they're published here. To run it yourself:

llama-server -m peek-q4_k_m-v<N>.gguf --port 8088 --alias peek \
  --chat-template-kwargs '{"enable_thinking":false}' -c 4096 -np 1

Turn thinking off: Peek was trained to answer straight away.

Training code and data: https://github.com/cosmospan/squadcode (training/).

Downloads last month: -

GGUF

Model size

0.8B params

Architecture

qwen35

Hardware compatibility

4-bit

Inference Providers NEW

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

Model tree for cosmospan/peek

Base model

Qwen/Qwen3.5-0.8B-Base

Finetuned

Qwen/Qwen3.5-0.8B

Quantized

(403)

this model