Instructions to use SII-Enigma/Davinci-Tool-30B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use SII-Enigma/Davinci-Tool-30B with Transformers:

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

pipe = pipeline("text-generation", model="SII-Enigma/Davinci-Tool-30B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("SII-Enigma/Davinci-Tool-30B")
model = AutoModelForCausalLM.from_pretrained("SII-Enigma/Davinci-Tool-30B", 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=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use SII-Enigma/Davinci-Tool-30B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "SII-Enigma/Davinci-Tool-30B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "SII-Enigma/Davinci-Tool-30B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/SII-Enigma/Davinci-Tool-30B

SGLang

How to use SII-Enigma/Davinci-Tool-30B 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 "SII-Enigma/Davinci-Tool-30B" \
    --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": "SII-Enigma/Davinci-Tool-30B",
		"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 "SII-Enigma/Davinci-Tool-30B" \
        --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": "SII-Enigma/Davinci-Tool-30B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use SII-Enigma/Davinci-Tool-30B with Docker Model Runner:
```
docker model run hf.co/SII-Enigma/Davinci-Tool-30B
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

Davinci-Tool-30B

📄 Paper · 💻 Code · 🤗 Davinci-Tool-30B · 🤗 Davinci-Tool-30B-SFT · 🤗 Davinci-ToolData

Davinci-Tool-30B is a 30B-parameter (3B activated) Mixture-of-Experts model for agentic, multi-turn tool use, trained with SFT on Davinci-ToolData (30,395 execution-verified trajectories spanning 2,260 invoked tools) followed by decision-level DPO. It improves Qwen3-Coder-30B-A3B-Instruct from 52.4 to 88.2 on τ²-bench, from 30.4 to 62.8 on BFCL v4 multi-turn, and from 16.9 to 56.2 on MCP-Atlas, outperforming Qwen3-Coder-480B-A35B-Instruct on all three benchmarks with 1/16 of its total parameters.

Benchmark comparison

Performance on τ²-bench, BFCL v4 multi-turn, and MCP-Atlas. Davinci-Tool-30B (after DPO) is hatched; its SFT checkpoint is shown for comparison.

Method

Davinci-Tool overview

Construction of Davinci-ToolData and training of Davinci-Tool.

Davinci-ToolData is built by two complementary pipelines that share one execution and verification interface:

EGTS — Execution-Grounded Task Synthesis (authored systems). An explorer interacts with a resettable executable world; user tasks are synthesized from the recorded outcome; the world is reset and the task is replayed with a guide, and the replay is accepted only if it recovers the explored outcome. Structural checks, a hard factual judge, and a soft quality judge filter the results.
EES — Executable Environment Synthesis (synthesized systems). Audited executable domains (191 domains in 19 categories) produce state-grounded tasks with executable verifiers within one domain (EES-Single), and are composed into shared multi-domain worlds (246 compositions) with shared entities, linking fields, cross-domain policies, and coherence constraints (EES-Cross).

Training has two stages: SFT on the verified demonstrations, then decision-level DPO, which reuses restorable states and executable verifiers to fork the SFT policy at decision points, execute alternative actions from the same state, and keep pairs whose verified downstream outcomes differ (1,150 pairs).

Main results

Model	τ²-bench Retail	τ²-bench Telecom	τ²-bench Avg.	BFCL v4 MT Avg.	MCP-Atlas
Qwen3-Coder-480B-A35B-Instruct	77.5	39.5	58.5	43.5	28.1
Kimi-K2-Instruct	70.6	65.8	68.2	33.4	59.6
MiniMax-M2.1	81.6	87.0	84.3	71.4	59.6
GLM-4.7	79.8	98.2	89.0	65.4	55.1
GPT-5.2	81.6	98.2	89.9	28.9	61.8
Claude-Sonnet-4.5	86.0	98.0	92.0	55.3	65.2
Qwen3-Coder-30B-A3B-Instruct	68.7	36.0	52.4	30.4	16.9
+ Toucan-1.5M SFT	65.8	89.5	77.7	41.9	22.5
+ Davinci-ToolData SFT (Davinci-Tool-30B-SFT)	73.7	96.5	85.1	60.6	52.8
+ SFT + DPO (Davinci-Tool-30B)	79.8	96.5	88.2	62.8	56.2

All scores are from the authors' evaluations under the official protocols; MCP-Atlas uses the 89 tasks that do not require external API keys. Under the same backbone and SFT hyperparameters, Davinci-ToolData exceeds the public Toucan-1.5M release by 7.4 / 18.7 / 30.3 points.

Transfer to smaller models (SFT on Davinci-ToolData):

Model	τ²-bench Avg.	BFCL v4 MT Avg.
Qwen3-8B → + Davinci-ToolData	32.5 → 74.6	15.4 → 33.9
Qwen3-14B → + Davinci-ToolData	33.4 → 76.8	21.1 → 41.4

Data sources are complementary (SFT on Qwen3-Coder-30B-A3B-Instruct):

EGTS	EES-Single	EES-Cross	τ²-bench	BFCL v4	MCP-Atlas
✓			78.9	57.4	28.1
	✓		82.0	54.5	29.2
		✓	83.4	46.6	47.2
✓	✓		79.4	57.6	31.5
	✓	✓	82.9	55.6	50.6
✓	✓	✓	85.1	60.6	52.8

Quickstart

The model uses the Qwen3-Coder chat template and tool-call format (chat_template.jinja).

vLLM

vllm serve SII-Enigma/Davinci-Tool-30B \
  --enable-auto-tool-choice \
  --tool-call-parser qwen3_coder \
  --max-model-len 65536

Then call it through any OpenAI-compatible client with a tools list.

Transformers

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "SII-Enigma/Davinci-Tool-30B"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype="auto", device_map="auto")

tools = [{
    "type": "function",
    "function": {
        "name": "get_weather",
        "description": "Get the current weather for a city.",
        "parameters": {
            "type": "object",
            "properties": {"city": {"type": "string"}},
            "required": ["city"],
        },
    },
}]
messages = [{"role": "user", "content": "What's the weather in Shanghai?"}]

inputs = tokenizer.apply_chat_template(
    messages, tools=tools, add_generation_prompt=True, return_tensors="pt"
).to(model.device)
outputs = model.generate(inputs, max_new_tokens=512, use_cache=True)
print(tokenizer.decode(outputs[0][inputs.shape[-1]:], skip_special_tokens=False))

generation_config.json ships the Qwen3-Coder default sampling settings (temperature=0.7, top_p=0.8, top_k=20, repetition_penalty=1.05).

Model details


Architecture	`Qwen3MoeForCausalLM`
Parameters	30.5B total, 3.3B activated
Layers / experts	48 layers, 128 experts, 8 activated per token
Attention	32 query heads, 4 KV heads (GQA)
Context (config)	262,144 tokens
Precision	BF16 safetensors

Training settings

Setting	SFT	DPO
Initialization	Qwen3-Coder-30B-A3B-Instruct	SFT checkpoint (also the frozen reference)
Data	Davinci-ToolData, 30,395 trajectories	1,150 decision-level preference pairs
Learning rate / schedule	1e-5, cosine, no warmup	5e-6, cosine, 3% warmup
Epochs	4	1
Max sequence length	32,768	24,576
Batch per device × grad. accum.	1 × 4	1 × 4
Other	assistant-token loss, sample packing, BF16, ZeRO-3	β = 1.0, precomputed reference log-probs

License

Released under the Apache 2.0 License, following the base model Qwen3-Coder-30B-A3B-Instruct.

Citation

@misc{yuan2026davincitoolscalingagentictooluse,
      title={Davinci-Tool: Scaling Agentic Tool-Use Supervision with Verifiable Executable Environments}, 
      author={Xiaoyang Yuan and Ji Zhao and Xingyu Hu and Wenhan Wang and Yutao Hu and Qiaosheng Zhang and Pengfei Liu and Yu Qiao and Yi Bin and Wenqi Shao},
      year={2026},
      eprint={2610.16867},
      archivePrefix={arXiv},
      primaryClass={cs.LG},
      url={https://arxiv.org/abs/2610.16867}, 
}