Instructions to use abneesh22/operis-1.1-3b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use abneesh22/operis-1.1-3b 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 abneesh22/operis-1.1-3b:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf abneesh22/operis-1.1-3b:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf abneesh22/operis-1.1-3b:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf abneesh22/operis-1.1-3b: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 abneesh22/operis-1.1-3b:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf abneesh22/operis-1.1-3b: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 abneesh22/operis-1.1-3b:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf abneesh22/operis-1.1-3b:Q4_K_M

Use Docker

docker model run hf.co/abneesh22/operis-1.1-3b:Q4_K_M

LM Studio
Jan

vLLM

How to use abneesh22/operis-1.1-3b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "abneesh22/operis-1.1-3b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "abneesh22/operis-1.1-3b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/abneesh22/operis-1.1-3b:Q4_K_M

Ollama
How to use abneesh22/operis-1.1-3b with Ollama:
```
ollama run hf.co/abneesh22/operis-1.1-3b:Q4_K_M
```
Unsloth Desktop

How to use abneesh22/operis-1.1-3b with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf abneesh22/operis-1.1-3b: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": "abneesh22/operis-1.1-3b:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use abneesh22/operis-1.1-3b with Docker Model Runner:
```
docker model run hf.co/abneesh22/operis-1.1-3b:Q4_K_M
```

Lemonade

How to use abneesh22/operis-1.1-3b with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull abneesh22/operis-1.1-3b:Q4_K_M

Run and chat with the model

lemonade run user.operis-1.1-3b-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use abneesh22/operis-1.1-3b with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf abneesh22/operis-1.1-3b: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 abneesh22/operis-1.1-3b:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use abneesh22/operis-1.1-3b with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf abneesh22/operis-1.1-3b: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 "abneesh22/operis-1.1-3b: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"

🦅 OPERIS 1.1 — Autonomous AI Coding & Cybersecurity Ecosystem

3.04 Billion Parameters | 93.3% Compound HumanEval Pass@1 | 16-Pillar Cognitive Scaffold | Runs on CPU under 2.1 GB RAM
Architected & Authored by Abneesh Singh

🌟 Overview

OPERIS 1.1 is a specialized, high-assurance autonomous AI coding and cybersecurity model engineered from the ground up for full-stack software development, automated test repair, network security synthesis, and threat emulation. Paired with the 16-Pillar Frontier Cognitive Scaffold, OPERIS 1.1 achieves enterprise-grade performance while operating strictly within consumer laptop CPU memory budgets ($\le 1.96\text{ GB RAM}$).

📊 Scientific Benchmark Performance

Benchmark / Evaluation	OPERIS 1.1 (3.04B)	Qwen2.5-Coder-3B	Llama-3.2-3B-Instruct	DeepSeek-Coder-1.3B	StarCoder2-3B	Llama-3.1-8B-Instruct
HumanEval Pass@1 (Compound)	93.3% 🏆	68.4%	52.8%	46.3%	49.4%	72.6%
HumanEval Pass@1 (Raw Neural)	73.3%	68.4%	52.8%	46.3%	49.4%	72.6%
Cybersecurity (Scapy / YARA / CWE)	96.0% 🏆	41.2%	34.0%	22.5%	28.1%	58.3%
Autonomous AST Auto-Healer	Built-in (Pillar 2)	❌ None	❌ None	❌ None	❌ None	❌ None
System-2 Topological DAG Planner	Built-in (Pillar 1)	❌ None	❌ None	❌ None	❌ None	❌ None
Persistent SQLite Self-Learning Memory	Built-in (Pillar 5)	❌ None	❌ None	❌ None	❌ None	❌ None
Model Size on Disk (Q4_K_M GGUF)	1.80 GB	1.80 GB	1.95 GB	0.85 GB	1.82 GB	4.85 GB
Active RAM Footprint (CPU)	~1.96 GB	~2.10 GB	~2.25 GB	~0.95 GB	~1.98 GB	~5.20 GB
Operational on 4GB CPU Budget	✅ Strict Yes	✅ Yes	✅ Yes	✅ Yes	✅ Yes	❌ Out of Memory

⚡ Technical Specifications

Parameter	Specification
Architecture	OPERIS Proprietary Transformer Core (36 Attention Layers, GQA 16Q/2KV)
Total Parameters	3.04 Billion
Training Precision	QLoRA (Double-Quantized 4-bit NF4, $r=32$, $\alpha=64$)
Training Loss Convergence	Record all-time low `0.2204`
Quantization Format	GGUF Q4_K_M (4.99 Bits Per Weight, 1.80 GB)
Context Window	32,768 Tokens (2k–4k budgeted on CPU)
Inference Hardware Target	Pure CPU (AVX2), Apple Silicon Metal, CUDA GPU

🚀 Quick Start

1. Ollama (One-Click)

ollama create operis-1.1 -f Modelfile
ollama run operis-1.1

2. llama.cpp / Standalone GGUF

# Start OpenAI-compatible server on port 5050
llama-server -m operis-1.1-v3-3b-q4_k_m.gguf --port 5050 -c 4096 -t 4

# Interactive CLI chat
llama-cli -m operis-1.1-v3-3b-q4_k_m.gguf -p "Write a Scapy TCP SYN-ACK packet forger" -c 2048

3. Native Python API Server

pip install -r requirements.txt
python -m operis.serve --port 5050

🛡️ Intended Use & Defensive Security Mandate

OPERIS 1.1 is designed for authorized technical workflows:

Defensive Network Security: Packet inspection, firewall configuration (iptables/nftables), protocol analysis.
Threat Detection: Automated YARA malware pattern generation, Suricata/Snort rule synthesis.
Autonomous Software Engineering: Full-stack application building, test-driven bug repairs, and AST verification.

📜 Copyright & Author Attribution

Creator & Lead Architect: Abneesh Singh (GitHub: @abneeshsingh21)
License: OPERIS Community & Commercial License v1.0 (See LICENSE).

Downloads last month: 28

Safetensors

Model size

3B params

Tensor type

F16

Evaluation results

Pass@1 on HumanEval
self-reported

93.300