Qwen3.7-Plus - A large-scale multimodal intelligent agent model launched by Alitongyi

What is Qwen3.7-Plus?

Qwen3.7-Plus is a new generation of multimodal large-scale model launched by Tongyi Qianwen, unifying vision and language into an integrated intelligent agent foundation. The model can perceive real-world scenes, read screens and operate GUIs, generate code based on visual references, support end-to-end navigation of mobile applications, answer visual questions by combining network knowledge, and seamlessly integrate GUI and CLI interactions within a single intelligent agent loop. As an all-around coding intelligent agent and productivity assistant, the model uses multimodal input to handle a full range of tasks from front-end prototyping to complex software engineering and multi-step workflow automation, and has cross-framework generalization capabilities.

Main functions of Qwen3.7-Plus

• Multimodal interactive hybrid agentIt unifies the processing of images, videos, screens, web pages, and text input, completing complex task loops within a GUI/CLI/tool environment.
• Visual intelligent agentsIt combines visual understanding, code interpreters, and search enhancement to solve visual puzzles, real-world question answering, and complex reasoning tasks.
• Visual programmingGenerate SVG, web pages, and interactive front-ends from images or videos, achieving end-to-end transformation from visual references to code.
• GUI intelligent agentUnderstand mobile and desktop interfaces, and perform control positioning, task planning, and multi-step operations.
• Real-world perception and reasoningIt covers real-world scenarios, document charts, OCR, videos, and driving scenario understanding.

Technical Principles of Qwen3.7-Plus

• Fusion of visual perception and reasoningThe model performs strongly on challenging visual reasoning benchmarks such as BabyVision, MathVision, and HiPhO, demonstrating a comprehensive understanding of image details, spatial relationships, physical common sense, and multi-step logic. In particular, it shows a significant improvement over its predecessor on BabyVision, indicating that the model has stronger generalization ability on tasks that more closely resemble early human visual cognition and spatial reasoning.
• End-to-end transformation from vision to codeThrough code interpreter integration, the model can transform visual problems into computable problem representations, autonomously write and execute code for solving, searching, or verifying. In tasks such as spot-the-difference, tile completion, Huarong Road, mazes, and jigsaw puzzles, the model can recognize image content, perform spatial modeling, path searching, state deduction, and result verification.
• GUI Automation and Multi-Step InteractionThe model can recognize screen content, locate key UI elements, understand task intent, and complete multi-step interactive operations. Significant improvements are achieved on ScreenSpot Pro, OSWorld-Verified, and AndroidWorld, supporting the transition from "understanding the interface" to "operating the interface" and "building the interface."
• Search-enhanced multimodal knowledge question answeringThe model combines visual input with external knowledge retrieval. It first extracts key entities, scenes, text, and contextual clues from the visual input, then obtains external knowledge through search, and finally provides an answer by combining visual evidence and retrieval results.
• Video understanding and driving scenario perceptionIt enhances the ability to process events, actions, temporal sequences, and semantic relationships in both short and long videos, while demonstrating a strong understanding of dynamic scenes, traffic participants, and spatial relationships in driving-related tests such as LingoQA, SURDS, and VLADBench.

How to use Qwen 3.7-Plus

• Visit the official platformAccess the model service via Alibaba Cloud Bailian or the Qwen Studio official website.
• Select model versionSelect Qwen3.7-Plus in the model marketplace and configure the call parameters according to your needs.
• Input multimodal contentIt supports uploading images, videos, screenshots, or web links, and interacting with them using text commands.
• Execute the taskSelect the appropriate capability mode (Visual Agent, GUI Agent, Visual Coding, etc.) based on the scenario, and the model will automatically complete the closed loop of perception, reasoning, and execution.

The core advantages of Qwen3.7-Plus

• Multimodal Agent Closed-Loop CapabilityIt integrates seeing, thinking, writing, doing, and verifying into a unified intelligent agent workflow, supporting the end-to-end automatic completion of complex software tasks from understanding to delivery.
• Cross-framework generalizationIt maintains stable performance regardless of whether it is deployed through Claude Code, OpenClaw, Qwen Code, or other frameworks.
• Leading in visual programmingIt scored 1772.0 on QwenVision2Code, close to GPT-5.4's 1884.0, and significantly outperformed Claude-Opus-4.6 (1518.0) and Gemini-3.1 Pro (1632.0).
• Strong GUI operation capabilitiesScreenSpot Pro 79.0 and AndroidWorld 81.0 are among the top tier in terms of interface understanding and operational tasks.
• Long-term autonomous operationThe case study shows that the Agent can run stably for 11+ hours, generating over 10,000+ lines of code and triggering over 1,000+ calls.

Qwen3.7-Plus project address

• Project official website: https://qwen.ai/blog?id=qwen3.7-plus

Comparison of Qwen3.7-Plus with similar competing products

Application Scenarios of Qwen3.7-Plus

• Intelligent software development: End-to-end APP development from requirements document generation to code writing, test case creation, GUI automated testing, and version iteration evolution.
• Desktop application replicaIt independently understands the UI layout and functional details of native applications, generates corresponding source code, and connects to real APIs to achieve high-fidelity application replication.
• Visual content generationTransform design reference images into executable SVG, web pages, or interactive front-end code, reducing the cost of visual-to-code assets.
• Multimodal knowledge Q&AIt combines images, videos, and web search to answer visual questions in the open world, such as location identification, event background analysis, and product information retrieval.
• Autonomous driving and embodied intelligenceUnderstanding dynamic driving scenarios, traffic participants, and spatial relationships supports real-world multimodal intelligent agents and embodied scenarios.