Instructions to use bg-digitalservices/Apertus-8B-2509-NVFP4A16 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use bg-digitalservices/Apertus-8B-2509-NVFP4A16 with Transformers:

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

pipe = pipeline("text-generation", model="bg-digitalservices/Apertus-8B-2509-NVFP4A16")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("bg-digitalservices/Apertus-8B-2509-NVFP4A16")
model = AutoModelForCausalLM.from_pretrained("bg-digitalservices/Apertus-8B-2509-NVFP4A16", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use bg-digitalservices/Apertus-8B-2509-NVFP4A16 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "bg-digitalservices/Apertus-8B-2509-NVFP4A16"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "bg-digitalservices/Apertus-8B-2509-NVFP4A16",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/bg-digitalservices/Apertus-8B-2509-NVFP4A16

SGLang

How to use bg-digitalservices/Apertus-8B-2509-NVFP4A16 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 "bg-digitalservices/Apertus-8B-2509-NVFP4A16" \
    --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": "bg-digitalservices/Apertus-8B-2509-NVFP4A16",
		"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 "bg-digitalservices/Apertus-8B-2509-NVFP4A16" \
        --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": "bg-digitalservices/Apertus-8B-2509-NVFP4A16",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use bg-digitalservices/Apertus-8B-2509-NVFP4A16 with Docker Model Runner:
```
docker model run hf.co/bg-digitalservices/Apertus-8B-2509-NVFP4A16
```

Apertus-8B-2509-NVFP4A16

NVFP4 quantization of swiss-ai/Apertus-8B-2509 — part of the Swiss AI Apertus model family. 8B dense transformer supporting 1,811 languages with 65K context.

W4A16 — weights in FP4, activations in FP16 (weight-only quantization). See also Apertus-8B-2509-NVFP4 for the full W4A4 variant.

Key Specs

	Original (BF16)	NVFP4 (this)
Size on disk	~16 GB	~5 GB
Compression	—	~3.0x
Parameters	8B	8B
Architecture	Dense transformer, xIELU activation	same
Context window	65,536 tokens	65,536 tokens
Languages	1,811	1,811

Serving with vLLM

vllm serve bg-digitalservices/Apertus-8B-2509-NVFP4A16 \
  --quantization modelopt \
  --dtype auto \
  --kv-cache-dtype fp8 \
  --gpu-memory-utilization 0.85 \
  --max-model-len 65536 \
  --trust-remote-code

DGX Spark

VLLM_NVFP4_GEMM_BACKEND=marlin vllm serve bg-digitalservices/Apertus-8B-2509-NVFP4A16 \
  --quantization modelopt \
  --dtype auto \
  --kv-cache-dtype fp8 \
  --max-model-len 65536 \
  --trust-remote-code

Testing

This is a base model.

Quantization Details

Method: NVIDIA Model Optimizer (modelopt) v0.43
Format: NVFP4 — E2M1 weights with per-group FP8 scales (group size 16)
Calibration: 4096 samples from CNN/DailyMail, batch size 32, seq_len 1024
Hardware: NVIDIA H200 GPU
Quantization script: included as quantize.py

About Apertus

Apertus is built by Swiss AI — a fully open, privacy-first model family trained on 4,096 GH200 GPUs. Key features:

1,811 native languages
Novel xIELU activation + AdEMAMix optimizer
EU AI Act compliant, respects opt-out consent
Full training transparency (weights, data, scripts all public)

License

Apache 2.0 — inherited from the base model.

Citation

If you use this model, please cite the original Apertus work:

@misc{swisstransformer2025apertus,
  title   = {Apertus},
  author  = {Swiss Transformer},
  year    = {2025},
  url     = {https://huggingface.co/swiss-ai}
}

Credits

Quantized by Mario Iseli on an NVIDIA H200. Built and validated with AI-engineering assistance from Anthropic.

📬 mario@marioiseli.com ☕ Buy me a coffee if this makes your inference go brrrrrr! 🚀

Downloads last month: 21

Safetensors

Model size

5B params

Tensor type

BF16

F8_E4M3

Model tree for bg-digitalservices/Apertus-8B-2509-NVFP4A16

Base model

swiss-ai/Apertus-8B-2509

Quantized

(13)

this model