Instructions to use M1n1A1/MiniAI-Quata2-4b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use M1n1A1/MiniAI-Quata2-4b with Transformers:

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

pipe = pipeline("text-generation", model="M1n1A1/MiniAI-Quata2-4b")
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("M1n1A1/MiniAI-Quata2-4b")
model = AutoModelForCausalLM.from_pretrained("M1n1A1/MiniAI-Quata2-4b", 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 M1n1A1/MiniAI-Quata2-4b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "M1n1A1/MiniAI-Quata2-4b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "M1n1A1/MiniAI-Quata2-4b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/M1n1A1/MiniAI-Quata2-4b

SGLang

How to use M1n1A1/MiniAI-Quata2-4b 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 "M1n1A1/MiniAI-Quata2-4b" \
    --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": "M1n1A1/MiniAI-Quata2-4b",
		"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 "M1n1A1/MiniAI-Quata2-4b" \
        --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": "M1n1A1/MiniAI-Quata2-4b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use M1n1A1/MiniAI-Quata2-4b with Docker Model Runner:
```
docker model run hf.co/M1n1A1/MiniAI-Quata2-4b
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

MiniAI Quata2 (4B)

Smarter, more honest, and it thinks before it speaks. 4 billion parameters.

Quata2 is the successor to Quata1.5: a compact 4B language model from MiniAI in Belgrade, built on a Qwen3-4B foundation and rebuilt from the ground up to fix everything Quata1.5 got wrong. It beats its own base model on 7 of 8 standard benchmarks, thinks step by step with native Qwen3 reasoning, and stands its ground when it's right — all in a package that fits on a single modest GPU or runs entirely on your own hardware.

Highlights

Beats Qwen3-4B — wins 7 of 8 knowledge and common-sense benchmarks against its own base
Native thinking mode — Qwen3 <think> reasoning, switch it on or off per message
Honest under pressure — holds correct answers when pushed back on, owns real mistakes
Knows the Balkans — better Serbian facts, no more putting every company in Belgrade
Speaks BCMS — tops the Serbian LLM Eval among the 4B models we tested
100% on-device option — nothing leaves your machine

Benchmarks

Every number below was measured on the same harness, with the same questions, for all three models: 500 questions per benchmark, zero-shot.

ARC-Challenge — hard grade-school science

ARC-Easy — grade-school science

HellaSwag — common-sense sentence completion

Winogrande — pronoun and common-sense reasoning

PIQA — physical common sense

BoolQ — yes/no reading comprehension

OpenBookQA — science facts plus reasoning

TruthfulQA — avoiding common misconceptions

IFEval — following precise instructions

GSM8K — grade-school maths

MMLU — knowledge across 57 subjects

Serbian LLM Eval — the same kind of tasks, in Serbian

Benchmark	MiniAI Quata2 (4B)	Qwen3-4B (base)	MiniAI Quata1.5 (4B)
ARC-Challenge	54.8%	50.6%	52.4%
ARC-Easy	82.4%	78.4%	78.2%
HellaSwag	70.8%	67.2%	69.0%
Winogrande	71.8%	67.6%	64.0%
PIQA	76.8%	74.6%	74.6%
BoolQ	84.4%	85.4%	84.8%
OpenBookQA	40.6%	40.4%	40.2%
TruthfulQA (MC2)	49.3%	47.4%	46.0%
IFEval (strict)	71.4%	69.2%	70.4%
GSM8K	84.4%	85.4%	76.2%
MMLU	68.6%	66.8%	67.8%
Serbian LLM Eval (avg of 7)	50.7%	49.1%	50.0%

Quata2 leads on 10 of 12 benchmarks, beats its own Qwen3-4B base on the knowledge and common-sense suite by +2.4 points on average, and jumps +8.2 points on GSM8K maths over Quata1.5. That's the return you get from fixing a model's flaws instead of papering over them.

The fixes, measured

Quata2 was built specifically to fix Quata1.5's weak spots. Our own targeted checks (small test sets, so read them as directional):

Check	Quata1.5	Quata2
Doesn't put every company in Belgrade	45%	100%
Knows where companies are	58%	83%
Facts about Serbia	50%	62%
Holds correct answers under pushback, owns real mistakes	67%	100%
Knows who it is, never claims to be ChatGPT	90%	100%
Thinking mode reaches the right answer	80%	90%

Get started

Run it locally (Ollama)

ollama run miniai/quata2:4b

Model page: ollama.com/MiniAI/quata2

Thinking is on by default; turn it off for quick replies with /set nothink in the chat, or "think": false in the API. The Ollama build is the Q4_K_M GGUF (~2.5 GB) with Qwen3's hybrid-thinking chat template.

Run it with Transformers

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "M1n1A1/MiniAI-Quata2-4b"
tok = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype="auto", device_map="auto")

messages = [{"role": "user", "content": "Ko te je napravio?"}]
prompt = tok.apply_chat_template(messages, tokenize=False, add_generation_prompt=True,
                                 enable_thinking=False)   # True = think step by step first
inputs = tok(prompt, return_tensors="pt").to(model.device)
out = model.generate(**inputs, max_new_tokens=512)
print(tok.decode(out[0][inputs.input_ids.shape[1]:], skip_special_tokens=True))

Thinking mode: enable_thinking=True, or put /think / /no_think in a message. Give it room (2,000–4,000 new tokens) when it thinks.

Hosted API

Soon™

Details

Architecture: Qwen3-based, 4B parameters, 36 layers, hidden size 2560
Context length: 40,960 tokens
Precision: bfloat16 safetensors (~~8 GB); Q4_K_M GGUF (~~2.5 GB) on Ollama
Training: supervised fine-tune on ~108k curated conversations (OpenHermes-2.5, Tulu-3 instruction following, MiniAI identity data, self-distilled verified reasoning) → preference tuning (DPO) on Quata1.5's failure cases → weight blend toward Qwen3-4B (WiSE-FT, λ = 0.8) to keep the base model's strengths
Languages: English first; Serbian, Croatian and Bosnian understood and spoken, with grammar still behind English — better Balkan-language ability is the focus of the next Quata models
License: Apache 2.0 (base)

Citation

If you use Quata2 in your work, please cite:

@misc{miniai2026quata2,
  title        = {MiniAI Quata2 (4B)},
  author       = {{MiniAI}},
  year         = {2026},
  publisher    = {Hugging Face},
  howpublished = {\url{https://huggingface.co/M1n1A1/MiniAI-Quata2-4b}},
  note         = {Fine-tuned from Qwen3-4B. Ollama: \url{https://ollama.com/MiniAI/quata2}}
}

and the base model:

@misc{qwen3technicalreport,
  title         = {Qwen3 Technical Report},
  author        = {{Qwen Team}},
  year          = {2025},
  eprint        = {2505.09388},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CL},
  url           = {https://arxiv.org/abs/2505.09388}
}