Quipu 0.8B
A compact, fast fine-tuned language model built on Qwen3.5-0.8B, tuned for clear step-by-step reasoning, honest tool-calling behavior, consistent identity, and lightweight coding assistance.
Designed to punch above its weight class: at just 0.8B parameters, Quipu runs fast and cheap while staying focused on giving structured, verified answers instead of confidently guessing — a solid pick when you need a responsive assistant without the overhead of a much larger model.
Good for
- Step-by-step reasoning and simple logic problems
- Basic coding help (short functions, quick snippets)
- Tool-calling / agentic workflows (web search, multi-step tasks)
- Light roleplay and character consistency
- Fast, low-resource deployment (edge devices, quick prototyping, local inference)
Model Details
- Base model: Qwen/Qwen3.5-0.8B
- Fine-tuning method: LoRA
- Languages: English, Spanish
- License: Apache 2.0
How to Get Started
from transformers import AutoModelForCausalLM, AutoTokenizer
model = AutoModelForCausalLM.from_pretrained("Quipuai/quipu-0.8b")
tokenizer = AutoTokenizer.from_pretrained("Quipuai/quipu-0.8b")
messages = [{"role": "user", "content": "Who are you?"}]
text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
inputs = tokenizer(text, return_tensors="pt")
output = model.generate(**inputs, max_new_tokens=100)
print(tokenizer.decode(output[0], skip_special_tokens=True))
GGUF / Ollama
Quantized GGUF versions (compatible with Ollama, llama.cpp, LM Studio, etc.) are available at:
To run with Ollama:
ollama run hf.co/Quipuai/quipu-0.8b-GGUF:Q4_K_M