Qwen/Qwen3.5-9B-Base

VISIONConcurrent Unit Cost:1Model Size:9BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Feb 26, 2026License:apache-2.0Architecture:Transformer0.1K Open Weights Featherless Exclusive Cold

Qwen/Qwen3.5-9B-Base is a 9 billion parameter causal language model developed by Qwen, featuring a unified vision-language foundation and an efficient hybrid architecture. It integrates multimodal learning and Gated Delta Networks, achieving cross-generational parity with Qwen3 models across various benchmarks. This model is designed for fine-tuning, in-context learning, and research, offering expanded support for 201 languages and a native context length of 262,144 tokens, extensible up to 1,010,000 tokens.

Loading preview...

Overview

Qwen3.5-9B-Base is a 9 billion parameter pre-trained causal language model with a vision encoder, developed by the Qwen Team. It represents a significant advancement in foundation models, integrating breakthroughs in multimodal learning, architectural efficiency, and scalable reinforcement learning. The model is primarily intended for fine-tuning, in-context learning experiments, and other research or development purposes, rather than direct interaction.

Key Capabilities & Features

  • 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: Utilizes Gated Delta Networks combined with 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: Supports 201 languages and dialects, enabling inclusive, worldwide deployment.
  • Extended Context Length: Natively supports 262,144 tokens, extensible up to 1,010,000 tokens.
  • Optimized for PEFT: Control tokens are trained to allow efficient LoRA-style PEFT, mitigating the need to finetune embeddings despite a larger vocabulary.

When to Use This Model

  • Fine-tuning: Ideal for adapting to specific downstream tasks.
  • Research & Development: Suitable for exploring advanced multimodal capabilities and architectural efficiencies.
  • Multilingual Applications: Excellent for projects requiring broad linguistic support and nuanced cultural understanding.
  • High-Throughput Inference: Benefits from its efficient hybrid architecture for applications demanding low latency and cost.