RedHatAI/Qwen3.8-27B

VISIONPricing:Input $1.6 / Cached $0.15 / Output $12Concurrent Unit Cost:2Model Size:27BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Aug 17, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

Qwen3.8-27B is a 27 billion parameter causal language model with a native vision encoder, developed by Qwen. It offers comprehensive improvements across coding, professional work, research, and long-horizon agentic tasks, with a native context length of 262,144 tokens extensible up to 1,000,000. This model excels in agent execution, multimodal understanding of images and videos, and flexible thinking control, making it suitable for complex, multi-step tasks.

Loading preview...

Qwen3.8-27B: Advanced Multimodal Agentic AI

Qwen3.8-27B is the latest and most capable generation in the Qwen open-model family, building upon the Qwen3.5 architecture. This 27 billion parameter causal language model features a native vision encoder, enabling it to understand both images and videos, from STEM diagrams to hour-scale video content. It boasts a native context length of 262,144 tokens, which can be extended up to 1,000,000 tokens using techniques like YaRN.

Key Capabilities

  • Enhanced Agent Execution: Demonstrates stronger autonomous planning and improved handling of environment feedback, leading to more reliable end-to-end task completion across various domains.
  • Multimodal Understanding: Natively supports image and video understanding, allowing for processing and reasoning over visual inputs.
  • Flexible Thinking Control: Features adjustable reasoning depth via reasoning_effort (xhigh, medium, low) and retains reasoning context from historical messages with preserve_thinking.
  • Comprehensive Performance: Shows substantial gains in coding, professional work, research, and long-horizon agentic tasks, outperforming previous Qwen versions and several other models in benchmarks like SWE-bench Pro (61.7), CoWorkBench (70.7), and OSWorld-Verified (84.3).

Ideal Use Cases

  • Complex Agentic Workflows: Suited for tasks requiring autonomous planning, environment interaction, and multi-step execution, such as software engineering (SWE-bench Pro: 61.7) and long-horizon office work (CoWorkBench: 70.7).
  • Multimodal Applications: Excellent for scenarios involving visual data, including computer use (OSWorld-Verified: 84.3), browser use (WebArena-Verified: 64.8), mobile use (AndroidWorld: 81.9), and visual problem-solving (MathVision with CI: 94.6).
  • High-Context Reasoning: Beneficial for applications requiring processing and generating responses over very long texts or conversations, leveraging its 262,144-token native context and 1M extensible context.

Finetunes
3 models