Instructions to use zenlm/zen3-omni with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use zenlm/zen3-omni with Transformers:

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

pipe = pipeline("text-generation", model="zenlm/zen3-omni")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("zenlm/zen3-omni")
model = AutoModelForMultimodalLM.from_pretrained("zenlm/zen3-omni", device_map="auto")
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/zen3-omni with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "zenlm/zen3-omni"
# 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/zen3-omni",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/zenlm/zen3-omni

SGLang

How to use zenlm/zen3-omni 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/zen3-omni" \
    --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/zen3-omni",
		"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/zen3-omni" \
        --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/zen3-omni",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use zenlm/zen3-omni with Docker Model Runner:
```
docker model run hf.co/zenlm/zen3-omni
```

zeekay commited on Jul 14

Commit

085a0b3

verified ·

1 Parent(s): 55ed0db

docs: honest base attribution / canonical-name note

Browse files

Files changed (1) hide show

README.md +10 -7

README.md CHANGED Viewed

@@ -1,6 +1,7 @@
 ---
 library_name: transformers
 pipeline_tag: text-generation
 tags:
   - text-generation
   - multimodal
@@ -15,22 +16,24 @@ language:
 - en
 ---
-# Zen3 Omni
-**Zen LM by Hanzo AI** — Multimodal model supporting text, image, audio, and video understanding. 202K context for complex analysis.
 ## Specs
 | Property | Value |
 |----------|-------|
-| Parameters | 1T MoE |
-| Context Length | 202K tokens |
-| Architecture | Zen MoDE (Mixture of Distilled Experts) |
 | Generation | Zen3 |
 ## API Access
-This model is served exclusively via the Hanzo AI API. Weight downloads are not available for this model tier.
 ```python
 from openai import OpenAI
@@ -47,7 +50,7 @@ Get your API key at [console.hanzo.ai](https://console.hanzo.ai) — $5 free cre
 ## License
-Apache 2.0
 ---

 ---
 library_name: transformers
 pipeline_tag: text-generation
+base_model: Qwen/Qwen3-Omni-30B-A3B-Instruct
 tags:
   - text-generation
   - multimodal
 - en
 ---
+# zen3-omni
+Multimodal model supporting text, image, audio, and video understanding.
+Repackaged from [Qwen/Qwen3-Omni-30B-A3B-Instruct](https://huggingface.co/Qwen/Qwen3-Omni-30B-A3B-Instruct) (apache-2.0, Alibaba Qwen). **Not trained from scratch** — a permissively-licensed redistribution for the OSS-clean Zen model line.
 ## Specs
 | Property | Value |
 |----------|-------|
+| Parameters | 30B total / 3B active (MoE) |
+| Architecture | Qwen3-Omni MoE (`Qwen3OmniMoeForConditionalGeneration`) |
+| Modality | text, image, audio, video |
 | Generation | Zen3 |
 ## API Access
+Also served via the Hanzo AI API:
 ```python
 from openai import OpenAI
 ## License
+`apache-2.0`. Upstream: **Qwen/Qwen3-Omni-30B-A3B-Instruct** by Alibaba Qwen. Upstream LICENSE/NOTICE retained in-repo.
 ---