Qwen3.5 Plus

Balanced flagship model with 1M context window, supporting text, image, and video inputs with MoE architecture.

Overview

Qwen3.5-Plus is a balanced flagship model supporting text, image, and video inputs. Its text performance matches Qwen3-Max while being faster and more cost-effective. Its multimodal capabilities significantly outperform the Qwen3-VL series. Based on a MoE (Mixture of Experts) architecture.

Key Specifications

Parameter	Value
Context Window	1,000,000 tokens (1M)
Max Output	8,192 tokens
Architecture	MoE (Mixture of Experts)
Input Modalities	Text, Image, Video
Output Modalities	Text

Capabilities

Text Generation — Flagship-level reasoning and generation
Code Generation — Strong multi-language programming
Vision — Image and video understanding (outperforms Qwen3-VL series)
Function Calling — Structured tool use
JSON Mode — Guaranteed valid JSON output
Thinking Mode — Deep reasoning (enabled by default)
Streaming — Real-time token delivery
Context Caching — Reduce cost for repeated prompts

Pricing

Type	Price
Input	$0.32 / 1M tokens
Output	$1.92 / 1M tokens
Cache Read	$0.032 / 1M tokens
Cache Write	$0.4 / 1M tokens

Usage Example

response = client.chat.completions.create(
    model="qwen3.5-plus",
    messages=[{"role": "user", "content": "Hello!"}],
)

Best Practices

Ideal for medium-to-high complexity tasks needing multimodal input
Use thinking mode for complex reasoning; non-thinking mode for faster responses
Leverage the 1M context window for long-document analysis
Prefer this over Qwen3-VL for vision tasks requiring strong text performance