Zen Eco 4B Instruct

Efficient instruction-following model. Part of the Zen Eco family.

Overview

Zen Eco 4B Instruct is based on Qwen/Qwen3-4B (Apache 2.0, Alibaba Qwen), a compact 4B dense instruction model, packaged for the OSS-clean Zen Eco line. Not trained from scratch.

Property	Value
Parameters	4B (dense)
Architecture	Qwen3 (Qwen3ForCausalLM)
Context	32K
License	Apache 2.0

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained("zenlm/zen-eco-instruct")
tokenizer = AutoTokenizer.from_pretrained("zenlm/zen-eco-instruct")

messages = [{"role": "user", "content": "Explain quantum computing in simple terms."}]
inputs = tokenizer.apply_chat_template(messages, return_tensors="pt")
output = model.generate(inputs, max_new_tokens=512)
print(tokenizer.decode(output[0], skip_special_tokens=True))

Upstream

Built on Qwen/Qwen3-4B by Alibaba Qwen, licensed under Apache 2.0. See NOTICE.

Related

• zen-eco — Base 4B model
• zen-eco-thinking — Chain-of-thought variant
• zen-eco-coder — Code variant
• Zen LM — Full model family

Apache 2.0 · Zen LM · Hanzo AI