Instructions to use continuedev/instinct with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use continuedev/instinct with Transformers:

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

pipe = pipeline("text-generation", model="continuedev/instinct", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("continuedev/instinct")
model = AutoModelForCausalLM.from_pretrained("continuedev/instinct", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.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(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Inference
Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use continuedev/instinct with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/continuedev/instinct

SGLang

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

Docker Model Runner
How to use continuedev/instinct with Docker Model Runner:
```
docker model run hf.co/continuedev/instinct
```

Adarsh-Iyer commited on Sep 3, 2025

Commit

fe52af9

verified ·

1 Parent(s): c9fe452

Update README.md

Browse files

Files changed (1) hide show

README.md +5 -1

README.md CHANGED Viewed

@@ -10,7 +10,11 @@ This repo contains the model weights for **Instinct**, [Continue](https://contin
 ## Serving the model
-There are many ways to plug a local model into Continue; we internally used an endpoint served by [SGLang](https://github.com/sgl-project/sglang), which is one of the options below. We observed no significant performance changes with fp8 quantization, so this may be used if desired.
 * SGLang: `python3 -m sglang.launch_server --model-path continuedev/instinct --load-format safetensors`
 * vLLM  : `vllm serve continuedev/instinct --served-model-name instinct --load-format safetensors`

 ## Serving the model
+We've released a [Q4_K_M GGUF quantization of Instinct](https://huggingface.co/continuedev/instinct-GGUF), for efficient local inference.
+[Ollama Instructions coming soon]
+Besides Ollama, there are many ways to plug a local model into Continue; we internally used an endpoint served by [SGLang](https://github.com/sgl-project/sglang), which is one of the options below. Quantizing for faster inference is also an option that worked well for us.
 * SGLang: `python3 -m sglang.launch_server --model-path continuedev/instinct --load-format safetensors`
 * vLLM  : `vllm serve continuedev/instinct --served-model-name instinct --load-format safetensors`