Instructions to use LLMWildling/gemma-4-opencoder-40b-a8b-bf16-v2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use LLMWildling/gemma-4-opencoder-40b-a8b-bf16-v2 with Transformers:

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

pipe = pipeline("text-generation", model="LLMWildling/gemma-4-opencoder-40b-a8b-bf16-v2")
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("LLMWildling/gemma-4-opencoder-40b-a8b-bf16-v2")
model = AutoModelForMultimodalLM.from_pretrained("LLMWildling/gemma-4-opencoder-40b-a8b-bf16-v2", 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 LLMWildling/gemma-4-opencoder-40b-a8b-bf16-v2 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "LLMWildling/gemma-4-opencoder-40b-a8b-bf16-v2"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "LLMWildling/gemma-4-opencoder-40b-a8b-bf16-v2",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/LLMWildling/gemma-4-opencoder-40b-a8b-bf16-v2

SGLang

How to use LLMWildling/gemma-4-opencoder-40b-a8b-bf16-v2 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 "LLMWildling/gemma-4-opencoder-40b-a8b-bf16-v2" \
    --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": "LLMWildling/gemma-4-opencoder-40b-a8b-bf16-v2",
		"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 "LLMWildling/gemma-4-opencoder-40b-a8b-bf16-v2" \
        --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": "LLMWildling/gemma-4-opencoder-40b-a8b-bf16-v2",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use LLMWildling/gemma-4-opencoder-40b-a8b-bf16-v2 with Docker Model Runner:
```
docker model run hf.co/LLMWildling/gemma-4-opencoder-40b-a8b-bf16-v2
```

Gemma 4 OpenCoder 40B-A8B BF16 v2

Gemma 4 OpenCoder 40B-A8B v2 is a continued post-trained Gemma 4 coding model focused on agentic software engineering, long-context reasoning, and reliable tool use. It is designed for OpenCode workflows including repository navigation, implementation, editing, debugging, review, and sustained multi-file work.

The checkpoint contains approximately 42.4B text parameters and activates approximately 8.1B text parameters per token. It retains Gemma 4's native 262,144-token context window.

This BF16 release is a self-contained checkpoint using the standard Gemma 4 architecture. It does not require an adapter, external router, or auxiliary model.

Intended use

Agentic software engineering
OpenCode coding workflows
Long-context codebase navigation
Reasoning-enabled tool use
Multi-file implementation and editing
Debugging and code review

Run with reasoning and automatic tool calling enabled, and preserve the checkpoint's included chat template.

Model format

Weight format: BF16 safetensors
Architecture: Gemma 4 Mixture-of-Experts
Total text parameters: approximately 42.4B
Active text parameters: approximately 8.1B per token
Maximum context: 262,144 tokens

vLLM

vllm serve LLMWildling/gemma-4-opencoder-40b-a8b-bf16-v2 \
  --tensor-parallel-size 2 \
  --reasoning-parser gemma4 \
  --tool-call-parser gemma4 \
  --enable-auto-tool-choice \
  --max-model-len 262144

Adjust tensor parallelism and context length for the available hardware. For OpenCode, select the served model and keep reasoning and tool calling enabled.

Downloads last month: 35

Safetensors

Model size

43B params

Tensor type

BF16

Model tree for LLMWildling/gemma-4-opencoder-40b-a8b-bf16-v2

Base model

google/gemma-4-26B-A4B

Finetuned

google/gemma-4-26B-A4B-it

Finetuned

(151)

this model