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

Libraries

How to use moondream/moondream-next with Transformers:

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

pipe = pipeline("text-generation", model="moondream/moondream-next", trust_remote_code=True)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("moondream/moondream-next", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use moondream/moondream-next with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "moondream/moondream-next"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "moondream/moondream-next",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/moondream/moondream-next

SGLang

How to use moondream/moondream-next 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 "moondream/moondream-next" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "moondream/moondream-next",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "moondream/moondream-next" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "moondream/moondream-next",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use moondream/moondream-next with Docker Model Runner:
```
docker model run hf.co/moondream/moondream-next
```

err805 commited on Jun 29

Commit

9b6830e

verified ·

1 Parent(s): 9671b8f

Update model manifest

Browse files

Files changed (1) hide show

manifest.json +20 -34

manifest.json CHANGED Viewed

@@ -1,36 +1,22 @@
 {
-  "base_model": "/root/.cache/huggingface/hub/models--moondream--moondream3-preview/snapshots/5112966d1a723413b1c9a1e8bea272b72e647b35/model_fp8.pt",
-  "lora_ckpt": "/workspace/evals/merged-opd-rl/source/multi-rl-from-opd-dual-r512-p64-step000004002-ckpt.pt",
-  "source_run": "multi-rl-from-opd-dual-r512-p64-4k-20260617-235617",
-  "source_step": 4002,
-  "output": "/workspace/evals/merged-opd-rl/output/moondream3-preview__multi-rl-from-opd-dual-r512-p64-4k-20260617-235617__step-000004002/model.pt",
-  "dense_layers": [
-    0,
-    1,
-    2,
-    3
-  ],
-  "moe_layers": [
-    4,
-    5,
-    6,
-    7,
-    8,
-    9,
-    10,
-    11,
-    12,
-    13,
-    14,
-    15,
-    16,
-    17,
-    18,
-    19,
-    20,
-    21,
-    22,
-    23
-  ],
-  "format": "md3_flat_bf16_moe_merged"
 }

 {
+  "default_file": "model_fp8.pt",
+  "default_repo": "moondream/moondream3-preview",
+  "default_size": 10505451019,
+  "expert_dtype": "torch.float8_e4m3fn",
+  "expert_tensors": 40,
+  "format": "md3_flat_fp8_moe_merged",
+  "non_scale_float_dtype": "torch.bfloat16",
+  "output_file": "/workspace/runs/mdnext-fp8-repair/model_fp8.pt",
+  "output_sha256": "e1acbd3f5eb8a1e74c7500b8a85051483de0406fa847eae35d53ce4032bb6ac7",
+  "output_size": 10505451019,
+  "scale_dtype": "torch.float32",
+  "scale_tensors": 40,
+  "size_delta_vs_default": 0,
+  "source_file": "model.pt",
+  "source_repo": "moondream/moondream-next",
+  "source_sha256_previous": null,
+  "zip_layout": {
+    "entries": 876,
+    "payload_size": 10505277740
+  }
 }