Tooony133/Qwen-3.8-27B-DinkyDoo

VISIONPricing:Input $1.6 / Cached $0.15 / Output $12Concurrent Unit Cost:2Model Size:27BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Aug 21, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

Tooony133/Qwen-3.8-27B-DinkyDoo is a 27 billion parameter causal language model from the Qwen3.8 family, fine-tuned for enhanced performance. This model is a native vision-language model capable of understanding images and videos, and features flexible thinking control for complex, multi-step agentic tasks. It excels in coding, professional work, research, and long-horizon agentic tasks, offering improved autonomous planning and handling of environment feedback.

Loading preview...

Qwen3.8-27B-DinkyDoo Overview

Tooony133/Qwen-3.8-27B-DinkyDoo is a 27 billion parameter model from the Qwen3.8 series, built upon the Qwen3.5 architectural foundation. This model is a native vision-language model, offering comprehensive understanding of images and videos, including STEM diagrams, documents, and hour-scale videos. It features a 32,768 token native context length, extensible up to 1,000,000 tokens using YaRN scaling techniques.

Key Capabilities

  • Enhanced Agent Execution: Demonstrates stronger autonomous planning and improved handling of environment feedback for more reliable end-to-end task completion.
  • Flexible Thinking Control: Includes a default 'thinking mode' that can be disabled, with adjustable reasoning depth (reasoning_effort) and retention of historical reasoning context (preserve_thinking).
  • Multimodal Understanding: Supports native image and video input, enabling analysis of visual data alongside text.
  • Improved Coding and Agentic Performance: Shows significant gains across various coding benchmarks (e.g., SWE-bench Pro, DeepSWE 1.1, QwenSWEBench) and agentic tasks (e.g., CoWorkBench, JobBench, Agents' Last Exam) compared to previous Qwen versions and other models.
  • Broad Compatibility: Designed for integration with popular inference frameworks like Hugging Face Transformers, vLLM, SGLang, and TokenSpeed.

Use Cases

This model is particularly well-suited for applications requiring advanced agentic capabilities, complex coding tasks, professional work automation, and research. Its multimodal understanding makes it ideal for scenarios involving visual data analysis, such as interpreting diagrams, documents, or video content within a task workflow. The flexible thinking control allows for fine-tuning the model's reasoning process to balance accuracy and speed for diverse applications.

Finetunes
1 models