Felldude/Qwen3.5-9B-Uncensored

VISIONConcurrent Unit Cost:1Model Size:9BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Jun 30, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

Felldude/Qwen3.5-9B-Uncensored is a 9 billion parameter language model based on the Qwen3.5 architecture, featuring a 32K context length. This model was specifically trained using Chain of Thought and shows a significant reduction in censored outputs compared to its untrained counterpart. It integrates a unified vision-language foundation and an efficient hybrid architecture for high-throughput inference, making it suitable for applications requiring uncensored and multimodal understanding.

Loading preview...

Felldude/Qwen3.5-9B-Uncensored Overview

Felldude/Qwen3.5-9B-Uncensored is a 9 billion parameter model built on the Qwen3.5 architecture, notable for its 32K token context length. A key differentiator is its training methodology, which exclusively utilized Chain of Thought and resulted in a massive increase in uncensored and unfiltered outputs, with only about 5% of image-related queries returning censored results compared to 98% in the untrained model.

Key Capabilities & Enhancements

  • Unified Vision-Language Foundation: Achieves cross-generational parity with Qwen3 and outperforms Qwen3-VL models across reasoning, coding, agents, and visual understanding benchmarks through early fusion training on multimodal tokens.
  • Efficient Hybrid Architecture: Incorporates Gated Delta Networks and sparse Mixture-of-Experts for high-throughput inference with minimal latency and cost.
  • Scalable RL Generalization: Features reinforcement learning scaled across million-agent environments for robust real-world adaptability.
  • Global Linguistic Coverage: Expanded support for 201 languages and dialects, enabling inclusive worldwide deployment.
  • Next-Generation Training Infrastructure: Achieves near-100% multimodal training efficiency compared to text-only training.

Good For

  • Applications requiring uncensored and unfiltered text generation.
  • Tasks benefiting from multimodal understanding and reasoning.
  • Use cases demanding high-throughput inference and efficient operation.
  • Global applications needing broad linguistic support.

Finetunes
1 models