JetBrains-Research/Qwen3-14B-am

Overview

JetBrains-Research/Qwen3-14B-am is a modified version of the Qwen3-14B model, developed by JetBrains Research. This 14.8 billion parameter causal language model introduces an assistant mask token to enhance the identification of assistant-generated content in its outputs. It maintains all the advanced capabilities of the original Qwen3 series, which are known for their comprehensive suite of dense and mixture-of-experts (MoE) models.

Key Capabilities & Differentiators

• Assistant Mask Token: A unique modification that allows for better identification of assistant-generated tokens, making it a drop-in replacement for the original Qwen3-14B with added utility.
• Dual Thinking Modes: Seamlessly switches between a 'thinking mode' for complex logical reasoning, math, and coding, and a 'non-thinking mode' for efficient, general-purpose dialogue. This can be controlled via enable_thinking parameter or dynamic /think and /no_think tags in prompts.
• Enhanced Reasoning: Demonstrates significant improvements in mathematics, code generation, and commonsense logical reasoning, surpassing previous Qwen models.
• Superior Human Alignment: Excels in creative writing, role-playing, multi-turn dialogues, and instruction following, providing a natural and engaging conversational experience.
• Agentic Expertise: Strong capabilities for tool integration, achieving leading performance among open-source models in complex agent-based tasks, especially when used with Qwen-Agent.
• Multilingual Support: Supports over 100 languages and dialects, offering robust multilingual instruction following and translation.
• Extended Context Length: Natively handles up to 32,768 tokens, with validated support for up to 131,072 tokens using YaRN scaling.

Use Cases

This model is ideal for applications requiring advanced reasoning, complex instruction following, and agentic workflows, particularly where clear distinction of assistant output is beneficial. Its dual-mode functionality makes it versatile for both highly analytical and general conversational tasks. Developers can leverage its multilingual support for global applications and its extended context for processing long documents or complex dialogues.