Overview
Nexa-AI-4B-Instruct is an instruction-tuned large language model built on top of Qwen/Qwen3-4B-Instruct-2507.
This project is jointly developed by:
Nexa AI focuses on delivering a capable multilingual AI assistant with strong performance in:
- General conversation
- Instruction following
- Coding
- Mathematics
- Logical reasoning
- Tool calling & AI agents
- Multilingual understanding (including English, Hindi, Chinese, and more)
Base Model
Base Model: Qwen/Qwen3-4B-Instruct-2507
We sincerely thank the Qwen Team for releasing the Qwen3 model family under the Apache 2.0 License, which made this project possible.
Developers
Project: Nexa AI
Developed by:
Model Details
- Model Name: Nexa-AI-4B-Instruct
- Base Model: Qwen/Qwen3-4B-Instruct-2507
- Architecture: Transformer Decoder
- Parameters: ~4 Billion
- Context Length: 262,144 Tokens (Inherited from the base model)
- License: Apache-2.0 (Base model license)
Features
- High-quality instruction following
- Coding assistance
- Mathematical reasoning
- Agent & tool calling support
- Multilingual capabilities
- Long-context understanding
- Fine-tuned alignment for helpful responses
Table with columns: GPT-4.1-nano-2025-04-14, Qwen3-30B-A3B Non-Thinking, Qwen3-4B Non-Thinking, Nexa-AI-4B-Instruct | GPT-4.1-nano-2025-04-14 | Qwen3-30B-A3B Non-Thinking | Qwen3-4B Non-Thinking | Nexa-AI-4B-Instruct |
|---|
| Knowledge | | | | |
| MMLU-Pro | 62.8 | 69.1 | 58.0 | 69.6 |
| MMLU-Redux | 80.2 | 84.1 | 77.3 | |
*: For reproducibility, we report the win rates evaluated by GPT-4.1.
© 2026 Neura Tech AI & Lumina AI. All rights reserved.