ZENLLC/ZEN-1
ZENLLC/ZEN-1 is a 4 billion parameter causal language model developed by Qwen, based on the Qwen3 architecture. This instruction-tuned model, also known as Qwen3-4B-Instruct-2507, features a native context length of 262,144 tokens and is optimized for general capabilities including instruction following, logical reasoning, text comprehension, mathematics, science, coding, and tool usage. It demonstrates significant improvements in long-tail knowledge coverage across multiple languages and excels in subjective, open-ended tasks, making it suitable for generating helpful and high-quality text.
Loading preview...
Model Overview
ZENLLC/ZEN-1, or Qwen3-4B-Instruct-2507, is a 4 billion parameter causal language model developed by Qwen. It is an updated version of the Qwen3-4B non-thinking mode, featuring a substantial native context length of 262,144 tokens. This model is specifically designed to operate in a non-thinking mode, meaning it does not generate <think></think> blocks in its output.
Key Enhancements & Capabilities
This iteration brings significant improvements across various domains:
- General Capabilities: Enhanced instruction following, logical reasoning, text comprehension, mathematics, science, and coding.
- Tool Usage: Improved capabilities for tool integration and agentic workflows, with recommendations for using Qwen-Agent.
- Knowledge Coverage: Substantial gains in long-tail knowledge across multiple languages.
- Alignment: Markedly better alignment with user preferences for subjective and open-ended tasks, leading to more helpful and higher-quality text generation.
- Long-Context Understanding: Enhanced performance in 256K long-context understanding.
Performance Highlights
Benchmarks indicate ZENLLC/ZEN-1 outperforms its predecessor and other models in its class across several metrics. Notably, it achieves 69.6 on MMLU-Pro, 47.4 on AIME25, 80.2 on ZebraLogic, and 83.5 on Creative Writing v3, showcasing strong performance in knowledge, reasoning, and creative tasks. It also demonstrates competitive results in coding and multilingualism.
Recommended Use Cases
ZENLLC/ZEN-1 is well-suited for applications requiring:
- Complex Instruction Following: Generating detailed and accurate responses based on intricate prompts.
- Advanced Reasoning: Tasks involving mathematical problems, scientific queries, and logical deductions.
- Creative Content Generation: Producing high-quality, open-ended text and creative writing.
- Agentic Workflows: Integrating with tools for enhanced functionality, particularly with the Qwen-Agent framework.
- Multilingual Applications: Leveraging its improved long-tail knowledge and multilingual capabilities.