zenlm/zen-eco-instruct

TEXT GENERATIONConcurrent Unit Cost:1Model Size:4BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Sep 30, 2025License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

The zenlm/zen-eco-instruct is a 4 billion parameter instruction-tuned causal language model, repackaged from Alibaba Qwen's Qwen3-4B. This model is part of the permissively-licensed Zen model line, offering a general-purpose foundation for various NLP tasks. It provides a robust base for applications requiring a compact yet capable instruction-following model.

Loading preview...

Overview

zenlm/zen-eco-instruct is an instruction-tuned, general-purpose language model with 4 billion parameters. It is repackaged from the Qwen/Qwen3-4B model developed by Alibaba Qwen, making it a permissively-licensed redistribution within the Zen model line. This model is not trained from scratch but serves as an OSS-clean foundation for various applications.

Key Characteristics

  • Architecture: Based on the Qwen3 (Qwen3ForCausalLM) architecture.
  • Parameter Count: Features 4 billion dense parameters, offering a balance between performance and computational efficiency.
  • Context Length: Supports a context length of 32768 tokens, enabling processing of longer inputs.
  • License: Distributed under the apache-2.0 license, inheriting its permissive terms from the upstream Qwen3-4B model.

Use Cases

This model is suitable for a wide range of general-purpose instruction-following tasks where a compact yet capable model is required. Its permissive license makes it an accessible choice for open-source projects and commercial applications.