Not-For-All-Audiences

hermes-lineage

mtp

coding

tool-calling

reasoning

roleplay

bf16

Model card Files Files and versions

xet

Community

Instructions to use oktayd/Qwen3.6-35B-v2-MoE-Ablit-Heretic-Uncensor-Hermes-MTP-Vision-FT with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

Transformers

How to use oktayd/Qwen3.6-35B-v2-MoE-Ablit-Heretic-Uncensor-Hermes-MTP-Vision-FT with Transformers:

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

pipe = pipeline("image-text-to-text", model="oktayd/Qwen3.6-35B-v2-MoE-Ablit-Heretic-Uncensor-Hermes-MTP-Vision-FT")
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("oktayd/Qwen3.6-35B-v2-MoE-Ablit-Heretic-Uncensor-Hermes-MTP-Vision-FT")
model = AutoModelForMultimodalLM.from_pretrained("oktayd/Qwen3.6-35B-v2-MoE-Ablit-Heretic-Uncensor-Hermes-MTP-Vision-FT", 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 oktayd/Qwen3.6-35B-v2-MoE-Ablit-Heretic-Uncensor-Hermes-MTP-Vision-FT with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "oktayd/Qwen3.6-35B-v2-MoE-Ablit-Heretic-Uncensor-Hermes-MTP-Vision-FT"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "oktayd/Qwen3.6-35B-v2-MoE-Ablit-Heretic-Uncensor-Hermes-MTP-Vision-FT",
		"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/oktayd/Qwen3.6-35B-v2-MoE-Ablit-Heretic-Uncensor-Hermes-MTP-Vision-FT

SGLang

How to use oktayd/Qwen3.6-35B-v2-MoE-Ablit-Heretic-Uncensor-Hermes-MTP-Vision-FT 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 "oktayd/Qwen3.6-35B-v2-MoE-Ablit-Heretic-Uncensor-Hermes-MTP-Vision-FT" \
    --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": "oktayd/Qwen3.6-35B-v2-MoE-Ablit-Heretic-Uncensor-Hermes-MTP-Vision-FT",
		"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 "oktayd/Qwen3.6-35B-v2-MoE-Ablit-Heretic-Uncensor-Hermes-MTP-Vision-FT" \
        --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": "oktayd/Qwen3.6-35B-v2-MoE-Ablit-Heretic-Uncensor-Hermes-MTP-Vision-FT",
		"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 oktayd/Qwen3.6-35B-v2-MoE-Ablit-Heretic-Uncensor-Hermes-MTP-Vision-FT with Docker Model Runner:
```
docker model run hf.co/oktayd/Qwen3.6-35B-v2-MoE-Ablit-Heretic-Uncensor-Hermes-MTP-Vision-FT
```

Qwen3.6-35B-v2-MoE-Ablit-Heretic-Uncensor-Hermes-MTP-Vision-FT / BENCHMARK-PLAN.md

oktayd

Publish Q36 v2 naming, complete edition guide and honest benchmark scope

a668ea0 verified 22 days ago

preview code

Raw

History Blame Contribute Delete

3.32 kB

Zusätzliche Benchmarks und Laptop-Vergleich

Alle Einträge sind registriert, nicht ausgeführt. Quelle: Qwen3.8-27B-Modellkarte, Revision 1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0. Veröffentlichte Zahlen sind kein direkter Vergleich zu unseren Kurztests.

Benchmark	Veröffentlichtes Qwen3.8-Ergebnis	Lokale Kurztest-Fälle (Vorschlag)	Voraussetzung
Terminal Bench 2.1 (Terminus)	73.0	5	version_and_harness_required
SWE-bench Pro	61.7	5	refined_tasks_Claude_Code_256K_temp1_top_p095
NL2Repo-Bench	42.3	5	Claude_Code_no_repository_download_reward_hacking
DeepSWE 1.1	42.2	5	Claude_Code_256K_temp1_top_p095
QwenSWEBench	79.0	offen – intern	in_house_avg3_8h_timeout_32768_output_256K_context
CoWorkBench	70.7	offen – intern	in_house_productivity_tasks
JobBench	33.4	5	dataset_and_harness_availability_unverified
Agents' Last Exam	Pass@1 20.4; Score 42.9	5	two_distinct_metrics
IFBench	79.5	20	not_IFEval
GPQA Diamond	89.2	20	existing_5_prompt_diagnostic_not_full_benchmark
HLE	30.8	10	publisher_GPT4o_judge_separate_judge_budget
LiveCodeBench v6	90.3	10	existing_LiveCodeBench_revision_must_be_reconciled
OSWorld-Verified	84.3	5	real_actions_not_static_screenshot_qa
WebArena-Verified	64.8	5	official_grader_OSWorld_scaffold
AndroidWorld	81.9	5	emulator_and_state_reset_required
RecreationBench	47.1	offen – intern	in_house_Ubuntu_macOS_Windows_Android_web
ClawEval-MM	Pass@3 57.4; Average 56.9	5	official_pass3_requires_three_trials; local_single_trial_not_pass3
SWE-MM	38.6	5	public_dev_split_with_Opus47_appendix8_3_modifications
Vision2Web	62.9	5	Claude_Code_judged_gpt5.4_2026_03_05
MathVision	Without CI 90.0; With CI 94.6	10	corrected_annotations_boxed_final_CI_strata_separate
BabyVision	Without CI 65.7; With CI 85.6	10	CI_strata_separate
CharXiv (RQ)	Without CI 83.7; With CI 90.2	10	corrected_annotations_CI_strata_separate
OmniDocBench 1.5	91.1	10	official_metric_not_generic_accuracy
RealWorldQA	85.9	20	dataset_revision_and_image_preprocessing_lock_required
ERQA	65.5	10	dataset_revision_and_official_scoring_required

Vergleichsregeln

IFBench ist nicht IFEval, MathVision nicht MathVista und CharXiv nicht ChartQA. Screenshot-Fragen ersetzen keinen OSWorld-/WebArena-Lauf. Die Qwen-Karte nutzt teils korrigierte Referenzen, andere Harnesses, 256K-Kontext, drei Versuche oder externe Richter. Solange diese Voraussetzungen nicht identisch sind, stehen Publisherwerte in einer getrennten Referenzspalte. Bei unbekannter Verfügbarkeit wird kein offizieller Adapter behauptet.

Laptop

4 Quants × 2 Backends × 2 Runtime-Profile = 16 Q36-Konfigurationen. Mit beiden Baselines wären es bis zu 48, aber nur wenn dieselben Quants verfügbar und mit dem Laptop kompatibel sind. Modellidentität und Runtime-Optimierung sind getrennte Achsen. Zunächst kleine Smoke-/Speichertests, dann eine feste Qualitätsstichprobe; Leistung dreimal warm messen. Die große Agentenmatrix ist nicht automatisch ein kostenloser Kurzlauf.