Instructions to use thinktecture/qwen3.5-4b-toolcalling-ft-nextera-q4_k_m with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use thinktecture/qwen3.5-4b-toolcalling-ft-nextera-q4_k_m with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="thinktecture/qwen3.5-4b-toolcalling-ft-nextera-q4_k_m",
	filename="qwen3.5-4b-toolcalling-ft-nextera-q4_k_m.gguf",
)

llm.create_chat_completion(
	messages = [
		{
			"role": "user",
			"content": "What is the capital of France?"
		}
	]
)

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use thinktecture/qwen3.5-4b-toolcalling-ft-nextera-q4_k_m with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf thinktecture/qwen3.5-4b-toolcalling-ft-nextera-q4_k_m:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf thinktecture/qwen3.5-4b-toolcalling-ft-nextera-q4_k_m:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf thinktecture/qwen3.5-4b-toolcalling-ft-nextera-q4_k_m:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf thinktecture/qwen3.5-4b-toolcalling-ft-nextera-q4_k_m:Q4_K_M

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf thinktecture/qwen3.5-4b-toolcalling-ft-nextera-q4_k_m:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf thinktecture/qwen3.5-4b-toolcalling-ft-nextera-q4_k_m:Q4_K_M

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf thinktecture/qwen3.5-4b-toolcalling-ft-nextera-q4_k_m:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf thinktecture/qwen3.5-4b-toolcalling-ft-nextera-q4_k_m:Q4_K_M

Use Docker

docker model run hf.co/thinktecture/qwen3.5-4b-toolcalling-ft-nextera-q4_k_m:Q4_K_M

LM Studio
Jan

vLLM

How to use thinktecture/qwen3.5-4b-toolcalling-ft-nextera-q4_k_m with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "thinktecture/qwen3.5-4b-toolcalling-ft-nextera-q4_k_m"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "thinktecture/qwen3.5-4b-toolcalling-ft-nextera-q4_k_m",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/thinktecture/qwen3.5-4b-toolcalling-ft-nextera-q4_k_m:Q4_K_M

Ollama
How to use thinktecture/qwen3.5-4b-toolcalling-ft-nextera-q4_k_m with Ollama:
```
ollama run hf.co/thinktecture/qwen3.5-4b-toolcalling-ft-nextera-q4_k_m:Q4_K_M
```

Unsloth Studio

How to use thinktecture/qwen3.5-4b-toolcalling-ft-nextera-q4_k_m with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for thinktecture/qwen3.5-4b-toolcalling-ft-nextera-q4_k_m to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for thinktecture/qwen3.5-4b-toolcalling-ft-nextera-q4_k_m to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for thinktecture/qwen3.5-4b-toolcalling-ft-nextera-q4_k_m to start chatting

How to use thinktecture/qwen3.5-4b-toolcalling-ft-nextera-q4_k_m with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf thinktecture/qwen3.5-4b-toolcalling-ft-nextera-q4_k_m:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @mariozechner/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "thinktecture/qwen3.5-4b-toolcalling-ft-nextera-q4_k_m:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use thinktecture/qwen3.5-4b-toolcalling-ft-nextera-q4_k_m with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf thinktecture/qwen3.5-4b-toolcalling-ft-nextera-q4_k_m:Q4_K_M

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default thinktecture/qwen3.5-4b-toolcalling-ft-nextera-q4_k_m:Q4_K_M

Run Hermes

hermes

Atomic Chat new
Docker Model Runner
How to use thinktecture/qwen3.5-4b-toolcalling-ft-nextera-q4_k_m with Docker Model Runner:
```
docker model run hf.co/thinktecture/qwen3.5-4b-toolcalling-ft-nextera-q4_k_m:Q4_K_M
```

Lemonade

How to use thinktecture/qwen3.5-4b-toolcalling-ft-nextera-q4_k_m with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull thinktecture/qwen3.5-4b-toolcalling-ft-nextera-q4_k_m:Q4_K_M

Run and chat with the model

lemonade run user.qwen3.5-4b-toolcalling-ft-nextera-q4_k_m-Q4_K_M

List all available models

lemonade list

⚠️ Conference talk demo — not production weights.

This model accompanies a conference keynote on local on-device AI. Published as a reference for the fine-tuning patterns shown on stage — not a deployable artefact. No security audit, no SLA, pinned to the talk's state.

Source repository: thinktecture-labs/local-multi-model-agent-slm
Threat model + out-of-scope: SECURITY.md
Licensing details: MODEL_LICENSES.md
All five models in the stack: Collection — Local Multi-Model Agent — nextera fine-tunes

Qwen3.5-4B FT (f16) — Tool Calling


Base model	`Qwen/Qwen3.5-4B` (4.0B params)
License	Tongyi Qianwen License — see `MODEL_LICENSES.md`
Training script	`finetune/train_qwen35_toolcalling.py`
Method	QLoRA r=16, α=16, 2 epochs, lr=2e-4 (via Unsloth — CUDA only)
Training data	`data/training-data/qwen35_toolcalling_{scenario}.jsonl` (~1,300 hand-curated tool-call examples)
Hardware	CUDA required (Unsloth dependency). Tested on RTX PRO 6000.
Intended use	Tool selection (sql_query / calculator) + argument generation. Native OpenAI tool-calling format. `enable_thinking=False` to keep output clean for llama.cpp's autoparser.
Out of scope	Free-form chat, RAG synthesis, intent classification. The model is trained only on tool-call outputs.
Reference eval (Nextera, v9 post-2026-05-15 retrain)	Tool routing: 99.4%. Multi-step decomposition (gemma3-ft): 98.8%. Multi-step chain shape: 97.5% (78/80, deterministic — verified byte-identical across 3 runs). SQL exec validity: 100% (79/79). Calculator expression correctness: 95.0%.
Known failure modes	Occasionally generates `<think>` blocks despite `enable_thinking=false` — the `_strip_thinking` filter in `src/engine/inference/client.py` handles this at parse time. Will refuse to answer if the query is clearly outside both tools (correct behaviour, but eval treats as "wrong tool").

Downloads last month: 81

GGUF

Model size

4B params

Architecture

qwen35

Hardware compatibility

4-bit

Model tree for thinktecture/qwen3.5-4b-toolcalling-ft-nextera-q4_k_m

Base model

Qwen/Qwen3.5-4B-Base

Finetuned

Qwen/Qwen3.5-4B

Quantized

(262)

this model