Qwen3.6-Plus

Multimodal agentic coding model with 1M context and visual reasoning

Try Now

read docs

About model

Qwen3.6-Plus is Qwen's multimodal agentic model built on a hybrid architecture combining efficient linear attention with sparse MoE routing and a 1M token context window. It delivers strong agentic coding performance with 78.8% SWE-Bench Verified and 61.6 Terminal-Bench 2.0, alongside 90.4% GPQA Diamond for reasoning and 86.0% MMMU for multimodal understanding. The model supports visual coding from UI screenshots to production code, video understanding, and thinking mode with preserve_thinking for maintaining reasoning context across multi-turn agent sessions.

GPQA Diamond

90.40%

Scientific and technical reasoning

SWE-Bench Verified

78.80%

Repository-level autonomous coding

Terminal-Bench 2.0

61.6

Terminal automation and agentic coding

Model key capabilities

Agentic Coding: 78.8% SWE-Bench Verified and 61.6 Terminal-Bench 2.0 with strong frontend development for 3D scenes, games, and interactive web applications
Visual Coding & Multimodal: Generate production code from UI screenshots and design mockups, with 91.2 OmniDocBench for document understanding and 86.0% MMMU for multimodal reasoning
Reasoning with Thinking Mode: 90.4% GPQA Diamond with preserve_thinking for maintaining full reasoning context across multi-turn agentic sessions
1M Context: Process entire codebases, long documents, and complex multi-turn workflows on a hybrid linear attention + sparse MoE architecture

Performance benchmarks

Model	GPQA Diamond	HLE	LiveCodeBench	MATH500	SWE-bench verified
Qwen3.6-Plus	90.40%		87.10%		78.80%
Related open-source models
Competitor closed-source models
Claude Opus 4.6	90.5%	34.2%			78.7%
OpenAI o3	83.3%	24.9%		99.2%	62.3%
OpenAI o1	76.8%			96.4%	48.9%
GPT-4o	49.2%	2.7%	32.3%	89.3%	31.0%

API usage

cURL
Python
Typescript

Endpoint:

Qwen/Qwen3.6-Plus

curl -X POST "https://api.together.xyz/v1/chat/completions" \
  -H "Authorization: Bearer $TOGETHER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "Qwen/Qwen3.6-Plus",
    "messages": [
      {
        "role": "user",
        "content": "What are some fun things to do in New York?"
      }
    ]
}'

from together import Together

client = Together()

response = client.chat.completions.create(
  model="Qwen/Qwen3.6-Plus",
  messages=[
    {
      "role": "user",
      "content": "What are some fun things to do in New York?"
    }
  ]
)
print(response.choices[0].message.content)

import Together from 'together-ai';
const together = new Together();

const completion = await together.chat.completions.create({
  model: 'Qwen/Qwen3.6-Plus',
  messages: [
    {
      role: 'user',
      content: 'What are some fun things to do in New York?'
     }
  ],
});

console.log(completion.choices[0].message.content);

Model card
Architecture Overview:
• Hybrid architecture combining efficient linear attention with sparse mixture-of-experts routing
• 1M token context window by default
• Multimodal: text, image, and video input with visual coding and visual agent capabilities
• Thinking mode with preserve_thinking option for maintaining full reasoning context across multi-turn agentic tasks
• Reduces redundant reasoning and overall token consumption in agent scenarios when preserve_thinking is enabled

Training Methodology:
• Built on the Qwen3.5-Plus foundation with targeted improvements for agentic coding, frontend development, and multimodal perception
• Deep integration of reasoning, memory, and execution capabilities for agentic workflows
• Optimized for complex frontend development including 3D scenes, games, and interactive web applications
• Enhanced multimodal training for document understanding, chart parsing, UI understanding, and video reasoning

Performance Characteristics:
• 90.4% GPQA Diamond for scientific reasoning
• 87.1% LiveCodeBench v6 for code generation
• 78.8% SWE-Bench Verified for autonomous software engineering
• 61.6 Terminal-Bench 2.0 for terminal automation and agentic coding
• 91.2 OmniDocBench for document understanding
• 86.0% MMMU for multimodal understanding
• Strong multilingual performance across coding and reasoning benchmarks
‍
Prompting
Together AI API Access:
• Access Qwen3.6-Plus via Together AI APIs using the endpoint Qwen/Qwen3.6-Plus
• Authenticate using your Together AI API key in request headers
• Supports thinking mode with optional preserve_thinking for maintaining reasoning context across agent turns
• Multimodal input: text, image, and video
• $0.50 input / $3.00 output per million tokens
• Available on Together AI serverless infrastructure
‍
Applications & use cases
Agentic Coding:
• Repository-level problem solving with 78.8% SWE-Bench Verified
• Terminal automation and complex task execution with 61.6 Terminal-Bench 2.0
• Frontend web development: 3D scenes, games, interactive applications with strong visual fidelity
• Compatible with coding agent frameworks for production development workflows

Visual Coding & Multimodal Agents:
• Generate frontend code from UI screenshots, product prototypes, and design mockups
• Visual reasoning for document understanding (91.2 OmniDocBench), chart parsing, and UI analysis
• Video understanding with temporal reasoning across frames
• Perception-to-action loop for GUI agent scenarios

Reasoning & Long Context:
• 90.4% GPQA Diamond for scientific and technical reasoning
• Thinking mode with preserve_thinking for consistent decision-making across long agentic sessions
• 1M token context for processing entire codebases, long documents, and complex multi-turn workflows
• Multilingual reasoning and tool-calling across diverse domains
‍

Related models

Model specifications

Model data

Model provider
Qwen
Type
Reasoning
Vision
Video
Chat
Main use cases
Reasoning
Features
Function Calling
JSON Mode
Deployment
Serverless
Endpoint
Qwen/Qwen3.6-Plus
Context length
1M
Input price
$0.50 / 1M tokens
Output price
$3.00 / 1M tokens
Input modalities
Text
Image
Video
Output modalities
Text

Released
April 1, 2026
Category
Chat

Run in Playground

Quickstart docs

Deploy model

Qwen3.6-Plus

About model

API usage

Model card

Prompting

Applications & use cases