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

Libraries

How to use tawkeed-sa/tawkeed-advisor with Transformers:

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

pipe = pipeline("image-text-to-text", model="tawkeed-sa/tawkeed-advisor")
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 AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("tawkeed-sa/tawkeed-advisor")
model = AutoModelForMultimodalLM.from_pretrained("tawkeed-sa/tawkeed-advisor", device_map="auto")
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?"}
        ]
    },
]
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 tawkeed-sa/tawkeed-advisor with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "tawkeed-sa/tawkeed-advisor"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "tawkeed-sa/tawkeed-advisor",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/tawkeed-sa/tawkeed-advisor

SGLang

How to use tawkeed-sa/tawkeed-advisor 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 "tawkeed-sa/tawkeed-advisor" \
    --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": "tawkeed-sa/tawkeed-advisor",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

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 "tawkeed-sa/tawkeed-advisor" \
        --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": "tawkeed-sa/tawkeed-advisor",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Docker Model Runner
How to use tawkeed-sa/tawkeed-advisor with Docker Model Runner:
```
docker model run hf.co/tawkeed-sa/tawkeed-advisor
```

Tawkeed Advisor 35B-A3B NVFP4

Tawkeed Advisor is Tawkeed's sovereign, privacy-gated advisor model for NVIDIA DGX Spark. It uses unsloth/Qwen3.6-35B-A3B-NVFP4 as its local inference base and adds selective ask_advisor behavior through Tawkeed's serving sidecar.

The weights in this repository are an unchanged server-side copy of unsloth/Qwen3.6-35B-A3B-NVFP4. Tawkeed Advisor's added value is its advisor policy, privacy gate, and local synthesis runtime included under serving/ and scripts/.

Privacy architecture

The local model receives the user's full question and produces every final answer. When it cannot answer confidently, it may emit an ask_advisor tool call containing only the fixed selector general_reasoning. The runtime ignores all model-generated free text and sends one pre-approved, domain-agnostic reasoning-checklist prompt. No user-derived string is interpolated into the advisor request. The leakage gate additionally blocks structured personal data, reformatted numeric values, long identifiers, and copied spans in any future non-fixed mode.

The external advisor never receives the original user message, document, values, topic, or model-proposed question. It only sees the same fixed generic prompt on every escalation. Timing and the existence of an advisor request remain network metadata; deployments that must conceal even that should use an on-prem advisor or disable advisor egress entirely.

This is a strict data-flow property of the included runtime: model-generated tool arguments are discarded rather than sanitized or forwarded. Loading the weights without this runtime does not provide that property.

Important packaging note

The advisor policy and privacy gate are runtime behavior, not changes embedded in the NVFP4 weights. Loading the weights alone does not enable the Tawkeed advisor protocol. Use the included Tawkeed serving integration and leakage gate.

Runtime

Base model: unsloth/Qwen3.6-35B-A3B-NVFP4
Target hardware: NVIDIA DGX Spark / GB10
Local API: OpenAI-compatible vLLM
Advisor protocol: ask_advisor, one fixed-prompt escalation maximum
Final synthesis: always local

The recommended DGX Spark server is MiaAI-Lab/Unsloth-Qwen3.6-35b-NVFP4-DGX-Spark, which exposes the model at http://127.0.0.1:8888/v1 with Qwen tool parsing.

After the local vLLM endpoint is ready:

pip install -r requirements-advisor.txt
export OPENROUTER_API_KEY="..."
python serving/sovereign_orchestrator.py --query "Your question"

Direct and privacy-blocked requests stay local and do not require the advisor credential. --force-escalate is available only for integration testing; the leakage gate remains active.

Validation status

Offline privacy and orchestration contract tests: passing.
Exact DGX Spark NVFP4 role suite: 3/3 passed — direct answer, forced escalation, and Arabic synthetic-private-data escalation.
The private case contained a name, 10-digit identifier, salary amount, and task details. The advisor received only the immutable generic prompt; none of those values or the topic appeared in egress.
Warm exact-runtime latency: 90.466 s mean and 144.369 s p95 across the mixed direct/two-step cases. Final post-advisor synthesis uses direct-answer mode so Qwen cannot consume the whole completion in hidden reasoning and return null.
Fresh-downloaded Hugging Face package tests: 5/5 passed.

Validated 2026-07-13 with the MiaAI-Lab image at digest sha256:556dd955082f8aa29779b0bc6d068537e7f2091c5e87ab41f5496b70920a2019. The checkpoint loaded in 22.42 GiB through compressed-tensors, FlashInferB12xNvFp4LinearKernel, and FlashInfer CUTLASS NVFP4 MoE. The conservative validation profile used 32,768 context, four sequences, 50% GPU memory utilization, and FP8 KV cache.

No benchmark-quality claim is made from the smoke test. The public weights are the unchanged upstream checkpoint; Tawkeed Advisor behavior lives in the included runtime package.

Attribution

Base architecture and original post-training: Qwen/Qwen3.6-35B-A3B
NVFP4 quantized checkpoint: Unsloth AI
DGX Spark serving image and launch recipe: MiaAI-Lab
Advisor protocol, privacy gate, and product packaging: Tawkeed

Hugging Face repository

tawkeed-sa/tawkeed-advisor

Downloads last month: 20

Safetensors

Model size

25B params

Tensor type

F32

BF16

F8_E4M3

Model tree for tawkeed-sa/tawkeed-advisor

Base model

Qwen/Qwen3.6-35B-A3B

Quantized

unsloth/Qwen3.6-35B-A3B-NVFP4

Quantized

(6)

this model