Gael1125/Qwen3.6-27B-V5
Qwen3.6-27B-V5 is a 27 billion parameter causal language model with a vision encoder, developed by Qwen. It is optimized for agentic coding, handling frontend workflows and repository-level reasoning with enhanced fluency and precision. The model features thinking preservation, allowing it to retain reasoning context from historical messages for streamlined iterative development, and natively supports a context length of 262,144 tokens, extensible up to 1,010,000 tokens via YaRN scaling.
Loading preview...
Overview
Qwen3.6-27B is a 27 billion parameter causal language model with a vision encoder, developed by Qwen. It builds upon the Qwen3.5 series, prioritizing stability and real-world utility for developers. The model is designed to offer a more intuitive, responsive, and productive coding experience, supporting both pre-training and post-training stages.
Key Capabilities
- Agentic Coding: Significantly upgraded to handle frontend workflows and repository-level reasoning with improved fluency and precision. It excels in benchmarks like SWE-bench Verified (77.2), SWE-bench Pro (53.5), and Terminal-Bench 2.0 (59.3).
- Thinking Preservation: Introduces an option to retain reasoning context from historical messages, which streamlines iterative development, reduces overhead, and enhances decision consistency in agent scenarios.
- Multimodal Input: Supports text, image, and video inputs, making it a versatile tool for various applications, including visual question answering and document understanding.
- Extended Context Length: Natively supports a context length of 262,144 tokens, and can be extended up to 1,010,000 tokens using YaRN scaling techniques, crucial for long-horizon tasks.
- Strong Performance: Demonstrates competitive performance across various benchmarks, including MMLU-Pro (86.2), C-Eval (91.4), MMMU (82.9), and MathVista mini (87.4).
Good For
- Coding and Software Development: Ideal for developers requiring advanced agentic coding capabilities, including frontend development and repository-level code reasoning.
- Complex Problem Solving: Suitable for tasks requiring deep reasoning and knowledge, as evidenced by its strong performance in STEM & Reasoning benchmarks like GPQA Diamond (87.8) and AIME26 (94.1).
- Multimodal Applications: Excellent for applications involving image and video understanding, such as visual question answering, document analysis, and spatial intelligence tasks.
- Iterative Development Workflows: The thinking preservation feature makes it particularly useful for maintaining context in long, iterative development sessions or agent-based systems.