Qwen3.5 397B-A17B

qwen3.5-397b-a17b

The flagship model of the Qwen3.5 family — massive mixture-of-experts for code, reasoning and analysis.

CodeReasoningAnalysis

Technical Specifications

API Identifier	`qwen3.5-397b-a17b`
Plans	Pro
Maximum Context	262 144 tokens
Vision (images)	No
Native Web Access	No
Image Generation	No
Input Cost	$1.80 / M tokens
Output Cost	$7.20 / M tokens

Overview

Qwen3.5 397B-A17B is the most powerful model in Alibaba's new Qwen3.5 generation. With 397 billion total parameters and 17 billion activated per request thanks to its MoE architecture, it rivals the best models on the market for code, complex reasoning and in-depth analysis. Mingo reserves it for the Pro plan with 262K tokens of context, for the most demanding tasks.

Strengths

Qwen3.5 flagshipState-of-the-art codeComplex reasoningIn-depth analysis

Ideal Use Cases

Complex software architecture

Design and implement large-scale software systems.

Multi-step reasoning

Solve complex problems requiring multiple logical steps.

Advanced data analysis

Analyze large datasets and extract insights.

Critical code optimization

Optimize algorithms and code for performance.

Pricing

Input Tokens

$1.80

per million tokens

Output Tokens

$7.20

per million tokens

Related Models

Frequently Asked Questions

What does "397B-A17B" mean?

It means the model has 397 billion total parameters, of which 17 billion are activated per request thanks to its mixture-of-experts architecture.

Is Qwen3.5 397B-A17B reserved for Pro?

Yes, due to its performance and cost, Mingo reserves it for the Pro plan.

What are the competitors of Qwen3.5 397B-A17B?

It rivals models like DeepSeek R1 or Qwen3.7 Max for code, reasoning and analysis tasks.

Use Qwen3.5 397B-A17B with Mingo

Mingo routes your most demanding code, reasoning and analysis tasks to Qwen3.5 397B-A17B on the Pro plan.

Try Mingo for free