Qwen3-VL Technical Report

2025-11-26

AI Summary

This work introduces Qwen3-VL—the most capable vision-language model in the Qwen series—designed to unify text understanding, ultra-long-context modeling (up to 256K tokens), and multimodal reasoning across single/multiple images and video. Methodologically, it proposes an enhanced interleaved MRoPE positional encoding, a DeepStack cross-modal fusion architecture, and a fine-grained temporal alignment mechanism; integrates dense and Mixture-of-Experts (MoE) components; and employs multi-level ViT feature fusion with text-guided temporal modeling. Experiments demonstrate state-of-the-art performance on major multimodal benchmarks—including MMMU and MathVista—while significantly advancing long-document parsing, high-precision video temporal localization, and cross-modal referencing. Qwen3-VL establishes a robust foundation for multimodal agents, visual reasoning, and multimodal code generation.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Language Grounding & Multi-modal NLP

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

We introduce Qwen3-VL, the most capable vision-language model in the Qwen series to date, achieving superior performance across a broad range of multimodal benchmarks. It natively supports interleaved contexts of up to 256K tokens, seamlessly integrating text, images, and video. The model family includes both dense (2B/4B/8B/32B) and mixture-of-experts (30B-A3B/235B-A22B) variants to accommodate diverse latency-quality trade-offs. Qwen3-VL delivers three core pillars: (i) markedly stronger pure-text understanding, surpassing comparable text-only backbones in several cases; (ii) robust long-context comprehension with a native 256K-token window for both text and interleaved multimodal inputs, enabling faithful retention, retrieval, and cross-referencing across long documents and videos; and (iii) advanced multimodal reasoning across single-image, multi-image, and video tasks, demonstrating leading performance on comprehensive evaluations such as MMMU and visual-math benchmarks (e.g., MathVista and MathVision). Architecturally, we introduce three key upgrades: (i) an enhanced interleaved-MRoPE for stronger spatial-temporal modeling across images and video; (ii) DeepStack integration, which effectively leverages multi-level ViT features to tighten vision-language alignment; and (iii) text-based time alignment for video, evolving from T-RoPE to explicit textual timestamp alignment for more precise temporal grounding. Under comparable token budgets and latency constraints, Qwen3-VL achieves superior performance in both dense and Mixture-of-Experts (MoE) architectures. We envision Qwen3-VL serving as a foundational engine for image-grounded reasoning, agentic decision-making, and multimodal code intelligence in real-world workflows.

Problem

• Enhancing multimodal understanding across text, images, and video inputs
• Improving long-context comprehension for documents and videos up to 256K tokens
• Advancing reasoning capabilities for visual, mathematical, and cross-modal tasks

Innovation

• Enhanced interleaved-MRoPE for spatial-temporal modeling
• DeepStack integration for vision-language alignment
• Text-based time alignment for precise video grounding