Instructions to use zenlm/zen-omni-30b-instruct with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use zenlm/zen-omni-30b-instruct with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="zenlm/zen-omni-30b-instruct")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("zenlm/zen-omni-30b-instruct")
model = AutoModelForMultimodalLM.from_pretrained("zenlm/zen-omni-30b-instruct")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = processor.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use zenlm/zen-omni-30b-instruct with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "zenlm/zen-omni-30b-instruct"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "zenlm/zen-omni-30b-instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/zenlm/zen-omni-30b-instruct

SGLang

How to use zenlm/zen-omni-30b-instruct with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "zenlm/zen-omni-30b-instruct" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "zenlm/zen-omni-30b-instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "zenlm/zen-omni-30b-instruct" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "zenlm/zen-omni-30b-instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use zenlm/zen-omni-30b-instruct with Docker Model Runner:
```
docker model run hf.co/zenlm/zen-omni-30b-instruct
```

Zen-Omni (30B)

Part of the Zen AI Model Family

Model Description

Parameters: 30B
Architecture: Zen
Specialization: Multimodal understanding & generation
Training: Multimodal training with vision-language pairs
Context: 32K-128K tokens
Thinking: Up to 256,000 tokens

Files in This Repository

This repository contains ALL formats and quantizations:

🔷 SafeTensors (Original)

model.safetensors - Full precision weights
config.json - Model configuration
tokenizer.json - Fast tokenizer

🟢 GGUF Quantized

zen-omni-30b-instruct-Q4_K_M.gguf - 4-bit (recommended)
zen-omni-30b-instruct-Q5_K_M.gguf - 5-bit (balanced)
zen-omni-30b-instruct-Q8_0.gguf - 8-bit (high quality)

🍎 MLX (Apple Silicon)

mlx-4bit/ - 4-bit quantized for M-series
mlx-8bit/ - 8-bit quantized for M-series

Performance

Benchmark	Score	Rank
MMLU	68.4%	Top 10%
GSM8K	71.2%	Top 15%
HumanEval	48.3%	Top 20%

Quick Start

Transformers

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained("zenlm/zen-omni-30b-instruct")
tokenizer = AutoTokenizer.from_pretrained("zenlm/zen-omni-30b-instruct")

# With thinking mode
messages = [{"role": "user", "content": "Your question here"}]
text = tokenizer.apply_chat_template(messages, enable_thinking=True)

GGUF with llama.cpp

./main -m zen-omni-30b-instruct-Q4_K_M.gguf -p "Your prompt" -n 512

MLX for Apple Silicon

from mlx_lm import load, generate
model, tokenizer = load("zenlm/zen-omni-30b-instruct")
response = generate(model, tokenizer, "Your prompt", max_tokens=200)

Unique Training Background

Multimodal training with vision-language pairs

This model was specifically optimized for multimodal understanding & generation with careful attention to:

Inference efficiency
Memory footprint
Quality preservation
Thinking capabilities

Part of the Zen Family • Collection • GitHub

Downloads last month: 2

Safetensors

Model size

35B params

Tensor type

BF16

Model tree for zenlm/zen-omni-30b-instruct

Unable to build the model tree, the base model loops to the model itself. Learn more.

Collection including zenlm/zen-omni-30b-instruct

Zen Omni

Collection

Omni multi-modal + flagship chat models (omni, nano, next, pro, max). • 6 items • Updated 11 days ago

Evaluation results

MMLU on MMLU
self-reported

0.684