Instructions to use ServiceNow-AI/SuperApriel-15B-Instruct with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use ServiceNow-AI/SuperApriel-15B-Instruct with Transformers:

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

pipe = pipeline("text-generation", model="ServiceNow-AI/SuperApriel-15B-Instruct", trust_remote_code=True)
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# Load model directly
from transformers import AutoModelForImageTextToText
model = AutoModelForImageTextToText.from_pretrained("ServiceNow-AI/SuperApriel-15B-Instruct", trust_remote_code=True, dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use ServiceNow-AI/SuperApriel-15B-Instruct with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "ServiceNow-AI/SuperApriel-15B-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": "ServiceNow-AI/SuperApriel-15B-Instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/ServiceNow-AI/SuperApriel-15B-Instruct

SGLang

How to use ServiceNow-AI/SuperApriel-15B-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 "ServiceNow-AI/SuperApriel-15B-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": "ServiceNow-AI/SuperApriel-15B-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 "ServiceNow-AI/SuperApriel-15B-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": "ServiceNow-AI/SuperApriel-15B-Instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use ServiceNow-AI/SuperApriel-15B-Instruct with Docker Model Runner:
```
docker model run hf.co/ServiceNow-AI/SuperApriel-15B-Instruct
```

denisko Claude Opus 4.6 commited on Apr 10

Commit

3f2bdcd

1 Parent(s): 5ef0391

Fix tilde rendering in vLLM section (avoid markdown strikethrough)

Browse files

Files changed (1) hide show

README.md +5 -5

README.md CHANGED Viewed

@@ -156,8 +156,8 @@ Recommended settings: temperature 0.6; increase `max_new_tokens` for complex rea
 Preset selection and throughput-optimized serving require the vLLM plugin from [Fast-LLM](https://github.com/ServiceNow/Fast-LLM). Two serving modes are available:
-- **Single-preset mode**: Only the weights for the selected mixer placement are loaded (~27 GiB for bf16). Unused mixer weights are never loaded, so the model fits comfortably on a single GPU with room for KV cache.
-- **Supernet mode**: All four mixer weights are loaded per layer (~46 GiB for bf16), enabling instant placement switching at runtime via `collective_rpc` (~3–20 ms per switch, no engine reload).
 ### Installation
@@ -254,9 +254,9 @@ llm.collective_rpc("set_layer_placements", args=(pattern,))
 | Setup | GPU Memory | KV Cache | Notes |
 |-------|-----------|----------|-------|
-| Single-preset, 1 GPU | ~27 GiB | ~39 GiB | Best for fixed deployments |
-| Supernet, 1 GPU (`enforce_eager`) | ~46 GiB | ~20 GiB | Runtime switching, lower KV capacity |
-| Supernet, 2 GPU (TP=2) | ~23 GiB/GPU | ~50 GiB/GPU | Full compile + CUDA graphs |
 ### Per-Request Preset Selection

 Preset selection and throughput-optimized serving require the vLLM plugin from [Fast-LLM](https://github.com/ServiceNow/Fast-LLM). Two serving modes are available:
+- **Single-preset mode**: Only the weights for the selected mixer placement are loaded (approx. 27 GiB in bf16). Unused mixer weights are never loaded, so the model fits comfortably on a single GPU with room for KV cache.
+- **Supernet mode**: All four mixer weights are loaded per layer (approx. 46 GiB in bf16), enabling instant placement switching at runtime via `collective_rpc` (3–20 ms per switch, no engine reload).
 ### Installation
 | Setup | GPU Memory | KV Cache | Notes |
 |-------|-----------|----------|-------|
+| Single-preset, 1 GPU | 27 GiB | 39 GiB | Best for fixed deployments |
+| Supernet, 1 GPU (`enforce_eager`) | 46 GiB | 20 GiB | Runtime switching, lower KV capacity |
+| Supernet, 2 GPU (TP=2) | 23 GiB/GPU | 50 GiB/GPU | Full compile + CUDA graphs |
 ### Per-Request Preset Selection