Instructions to use RunSLM-AI/SmolLM2-1.7B-Instruct-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 RunSLM-AI/SmolLM2-1.7B-Instruct-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 RunSLM-AI/SmolLM2-1.7B-Instruct-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf RunSLM-AI/SmolLM2-1.7B-Instruct-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 RunSLM-AI/SmolLM2-1.7B-Instruct-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf RunSLM-AI/SmolLM2-1.7B-Instruct-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 RunSLM-AI/SmolLM2-1.7B-Instruct-GGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf RunSLM-AI/SmolLM2-1.7B-Instruct-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 RunSLM-AI/SmolLM2-1.7B-Instruct-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf RunSLM-AI/SmolLM2-1.7B-Instruct-GGUF:Q4_K_M

Use Docker

docker model run hf.co/RunSLM-AI/SmolLM2-1.7B-Instruct-GGUF:Q4_K_M

LM Studio
Jan

vLLM

How to use RunSLM-AI/SmolLM2-1.7B-Instruct-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "RunSLM-AI/SmolLM2-1.7B-Instruct-GGUF"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "RunSLM-AI/SmolLM2-1.7B-Instruct-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/RunSLM-AI/SmolLM2-1.7B-Instruct-GGUF:Q4_K_M

Ollama
How to use RunSLM-AI/SmolLM2-1.7B-Instruct-GGUF with Ollama:
```
ollama run hf.co/RunSLM-AI/SmolLM2-1.7B-Instruct-GGUF:Q4_K_M
```
Unsloth Desktop
Docker Model Runner
How to use RunSLM-AI/SmolLM2-1.7B-Instruct-GGUF with Docker Model Runner:
```
docker model run hf.co/RunSLM-AI/SmolLM2-1.7B-Instruct-GGUF:Q4_K_M
```

Lemonade

How to use RunSLM-AI/SmolLM2-1.7B-Instruct-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull RunSLM-AI/SmolLM2-1.7B-Instruct-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.SmolLM2-1.7B-Instruct-GGUF-Q4_K_M

List all available models

lemonade list

Atomic Chat

SmolLM2-1.7B-Instruct (GGUF for RunSLM Mobile Workstation)

Official, production-calibrated Q4_K_M GGUF quantization of HuggingFaceTB/SmolLM2-1.7B-Instruct, purpose-built for physical, zero-cloud execution in the RunSLM AI mobile workstation on Apple Silicon (Metal UMA) and Android (ARM64-v8.2a+ / KleidiAI SIMD).

⚡ Hardware Footprint & Operational Profile

Metric	Target Specification
Model Architecture	LLaMA-based (SmolLM2)
Parameter Count	1.71 Billion
Quantization Method	`Q4_K_M` (4-bit Medium K-Quants)
Binary Artifact	`SmolLM2-1.7B-Instruct-Q4_K_M.gguf`
File Size	1.06 GB (1,142,482,944 bytes)
Runtime KV Cache Footprint	~144 MB (`Q8_0` Key / `Q4_0` Value at 4,096 tokens)
Recommended Device RAM	4 GB+ (iOS & Android)
Observed Throughput	40–55+ t/s (Apple Silicon Metal) · 18–26+ t/s (ARM KleidiAI SIMD)
Time-To-First-Token (TTFT)	< 350 ms on modern mobile SoCs

🛠️ Direct Model Download

You can fetch the weights directly using curl or Hugging Face CLI:

# Direct HTTP download
curl -L -o SmolLM2-1.7B-Instruct-Q4_K_M.gguf \
  https://huggingface.co/RunSLM-AI/SmolLM2-1.7B-Instruct-GGUF/resolve/main/SmolLM2-1.7B-Instruct-Q4_K_M.gguf

USB Sideloading (RunSLM iOS / Android)

iPhone / iPad: Connect via USB to Windows/Mac → Open Apple Devices / iTunes → File Sharing → RunSLM → Drag and drop SmolLM2-1.7B-Instruct-Q4_K_M.gguf.
Android / Tablets: Connect via USB → Transfer directly into Android/data/com.tking0000.runslm/files/models/.

💬 Prompt Template: ChatML

SmolLM2-1.7B-Instruct natively parses the standard ChatML format. Ensure your tokenizer or inference wrapper uses the following structure:

<|im_start|>system
You are RunSLM, a concise, sovereign on-device assistant.<|im_end|>
<|im_start|>user
What is the difference between CPU cache and main memory?<|im_end|>
<|im_start|>assistant

Stop Sequences

Configure your engine to break generation on:

<|im_end|>
<|endoftext|>

🎙️ Intended Mobile Workloads

Air-Gapped Sovereign Chat: Real-time general reasoning, code generation, and instruction following with zero server communication.
Dual-Persona AI Podcast Studio: Low-latency turn generation between dialectical Host and Expert personas with zero acoustic overlap.
Local Workspace Synthesis: Drafting calendar events, parsing offline documents, and generating Markdown notes directly into mobile storage.

📜 Licensing & Legal Attribution

Base Model: SmolLM2-1.7B-Instruct released by Hugging Face, Inc. under the Apache 2.0 License.
Quantization & Packaging: Released by RunSLM AI under the Apache 2.0 License.
Commercial Use: Permitted under Apache 2.0 terms. Please preserve original copyright and license notices when redistributing.

Downloads last month: 151

GGUF

Model size

2B params

Architecture

llama

Hardware compatibility

4-bit

Model tree for RunSLM-AI/SmolLM2-1.7B-Instruct-GGUF

Base model

HuggingFaceTB/SmolLM2-1.7B

Quantized

HuggingFaceTB/SmolLM2-1.7B-Instruct

Quantized

(120)

this model