Instructions to use Soofi-Project/Soofi-S-Base with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Soofi-Project/Soofi-S-Base with Transformers:

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

pipe = pipeline("text-generation", model="Soofi-Project/Soofi-S-Base")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("Soofi-Project/Soofi-S-Base", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Soofi-Project/Soofi-S-Base with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/Soofi-Project/Soofi-S-Base

SGLang

How to use Soofi-Project/Soofi-S-Base 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 "Soofi-Project/Soofi-S-Base" \
    --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": "Soofi-Project/Soofi-S-Base",
		"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 "Soofi-Project/Soofi-S-Base" \
        --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": "Soofi-Project/Soofi-S-Base",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Soofi-Project/Soofi-S-Base with Docker Model Runner:
```
docker model run hf.co/Soofi-Project/Soofi-S-Base
```

Access request pending since July 16

by MLueck84 - opened Aug 19

Discussion

MLueck84

Aug 19

Hi Soofi team,

my access request for the Soofi S Beta Models has been pending since July 16.

We would like to evaluate Soofi S internally for German/English enterprise AI agent workloads. No redistribution, evaluation only.

Could you please let me know whether beta access is still being reviewed?

Best regards
Marco

chrisfinger

Aug 24

same here

RudiTheRude

Aug 31

Same for me. When can we test it?

alimhanna91

Sep 11

for me from 27 Juli , with no response

kno10

22 days ago

•

edited 22 days ago

They will be approved in one year, when the model is obsolete, but the next grant has been approved.
See OpenGPT-X / Teuken. It also took a year to be actually available

SOOFI is a JOKE.

llmguy342

9 days ago

damn

M-vdBerg

4 days ago

For anyone still waiting here: the team sent a community mail on 1 October that answers part of this, though none of it has made it into the repo. The relevant points:

The checkpoints on HF are intermediate checkpoints, in a closed beta with already selected partners.
The final checkpoints will be freely available, without gated access.
The target is an open release this year, possibly extending into next year, depending on post-training and evaluation.

Read alongside this thread, the pending requests look less like a queue being worked through and more like a queue that was quietly closed.
If the beta cohort is fixed, saying so plainly would be more useful than leaving requests from mid-July "in review" — people are planning evaluations around access they may never get.

I'll say openly that the current state is disappointing. "This year, possibly next year" is a wide window for a model whose homepage still promises a release this year, and a date that keeps moving is hard to plan against. That isn't a verdict on the work: the technical report and the public training and eval code are solid and useful. It's about the communication around them.

Two concrete requests:

Say whether beta access requests are still being reviewed at all, or whether the cohort is closed. Either answer works; no answer doesn't.
Mirror the status from the community mail — and the license intended for the final checkpoints — on the model card. The people affected are reading the repository, not an inbox.

Best,
Michael

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment