Qwen 3.5 9B

Alibaba

Most popular local Qwen 3.5 size. Hybrid architecture with Gated Delta Networks and sparse MoE. Supports 201 languages, thinking/non-thinking modes, and native multimodal (text + image + video).

Text Generation Local Qwen Family v3.5
Documentation Back to Models
Parameters
9B
params
Context Window
262K
tokens
Max Output
66K
tokens
Input Price
-
per 1M tokens
Output Price
-
per 1M tokens

Qwen Family Timeline 13 versions

Qwen 3.5 35B-A3B local
Feb 2026
Qwen 3.5 27B Latest local
Feb 2026
Qwen 3.5 9B Current local
Feb 2026
Qwen 3.5 Plus Latest api
Feb 2026
Qwen 3 Max api
Jul 2025
Qwen 3 MoE Small api
Apr 2025
Qwen 3 MoE Large api
Apr 2025
Qwen 2.5 Max api
Jan 2025
QwQ 32B (Reasoning) local
Nov 2024
Qwen 2.5 7B local
Sep 2024
Qwen 2.5 14B local
Sep 2024
Qwen 2.5 32B local
Sep 2024
Qwen 2.5 Coder 32B local
Sep 2024

Capabilities

👁️
Vision
⚡
Function Calling
📋
JSON Mode
🌊
Streaming
💬
System Prompt
🖥️
Code Execution
🔍
Web Search
🔌
MCP Support

Local Model Specs

Quantization
Q4_K_M
Architecture
Hybrid MoE (Gated Delta Networks)
Runtime
Ollama / llama.cpp
Disk Size
6.6 GB

Details

Release Date
February 16, 2026
Knowledge Cutoff
-
Source
Local
License
Apache 2.0
Model ID
qwen3.5-9b
Last updated: March 13, 2026