Instructions to use bottlecapai/ThinkingCap-Qwen3.8-27B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use bottlecapai/ThinkingCap-Qwen3.8-27B with Transformers:

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

pipe = pipeline("image-text-to-text", model="bottlecapai/ThinkingCap-Qwen3.8-27B")
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)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("bottlecapai/ThinkingCap-Qwen3.8-27B")
model = AutoModelForMultimodalLM.from_pretrained("bottlecapai/ThinkingCap-Qwen3.8-27B", 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=256)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use bottlecapai/ThinkingCap-Qwen3.8-27B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "bottlecapai/ThinkingCap-Qwen3.8-27B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "bottlecapai/ThinkingCap-Qwen3.8-27B",
		"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/bottlecapai/ThinkingCap-Qwen3.8-27B

SGLang

How to use bottlecapai/ThinkingCap-Qwen3.8-27B 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 "bottlecapai/ThinkingCap-Qwen3.8-27B" \
    --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": "bottlecapai/ThinkingCap-Qwen3.8-27B",
		"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 "bottlecapai/ThinkingCap-Qwen3.8-27B" \
        --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": "bottlecapai/ThinkingCap-Qwen3.8-27B",
		"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 bottlecapai/ThinkingCap-Qwen3.8-27B with Docker Model Runner:
```
docker model run hf.co/bottlecapai/ThinkingCap-Qwen3.8-27B
```

Could ThinkingCap be combined with Amazon LAR-27B for stronger agentic coding?

by Schestex - opened 17 days ago

Discussion

Schestex

17 days ago

Hi BottleCap team,

first of all, thank you for ThinkingCap-Qwen3.8-27B!

The reduction in reasoning tokens while preserving most of the original model's capabilities is particularly valuable for local coding agents.

I recently came across Amazon's new LAR-27B (Lean4AgenticRL):

https://huggingface.co/amazon/LAR-27B

Amazon reports some very interesting agentic coding results:

SWE-bench Verified: 84.2%
Terminal-Bench 2.0: 68.2%
Terminal-Bench 2.1: 77.3%

LAR-27B uses multi-level process rewards to improve agentic software-engineering behavior, including tool use and multi-step problem-solving.
Since both LAR-27B and ThinkingCap are derived from Qwen3.8-27B, I was wondering whether you would consider combining their strengths.

The ideal result would be a model that preserves:

ThinkingCap's efficient, shorter reasoning traces.
LAR-27B's improved agentic coding and tool-use capabilities.

Strong instruction following and long-context performance.
Compatibility with Qwen3.8's native MTP speculative decoding.

A few technical questions:

Do you think a model merge (e.g., TIES/DARE or another weight-merging approach) could work, or would sequential fine-tuning, distillation, or additional RL be necessary?

Could LAR's agentic capabilities be transferred into ThinkingCap without losing its reasoning-token efficiency?
Would it be feasible to preserve or recover native MTP acceptance after combining the models?

Would you be interested in experimenting with such a combination and evaluating it against both original checkpoints?

My practical use case is running ThinkingCap locally through NInfer on a 24 GB NVIDIA Blackwell GPU, primarily for coding agents, repository modifications, and long tool-use loops.

I believe a ThinkingCap + LAR agentic variant could be a compelling combination of efficient reasoning and strong software-engineering capabilities.

Would this be technically feasible, and is it something your team might consider exploring?

Thanks again for your work!

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