The Sequence AI of the Week #818: You Cannot Miss Qwen 3.5

2026-03-04 · Source: TheSequence · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Emerging Technologies & Innovation · Depth: Advanced, quick

Summary

The Alibaba Qwen team recently launched the Qwen 3.5 series, featuring a flagship Qwen3.5-397B-A17B model that competes with proprietary giants like GPT-5.2 and Claude Opus 4.5 on benchmarks. This release includes "Medium" models such as Qwen3.5-35B-A3B and "Small" models ranging from 0.8B to 9B parameters, specifically designed for on-device edge computing. Beyond mere scale, Qwen 3.5 represents a significant architectural pivot, abandoning pure dense transformers for reimagined attention mechanisms and extreme Mixture-of-Experts (MoE) sparsity. A key innovation is its native multimodality, even in sizes small enough to run on a smartphone, marking it as a notable engineering achievement in the open-weight AI ecosystem.

Key takeaway

Qwen 3.5 introduces a profound architectural pivot, abandoning pure dense transformers for extreme Mixture-of-Experts (MoE) sparsity, reimagined attention, and native multimodality. Its flagship models (e.g., 397B-A17B) benchmark competitively against proprietary giants like GPT-5.2 and Claude Opus 4.5. This enables advanced AI from high-performance cloud deployments to efficient on-device edge computing with models as small as 0.8B parameters.

Topics

• Qwen 3.5
• Mixture-of-Experts
• Multimodal AI
• Edge AI
• Transformer Architecture

Best for: CTO, VP of Engineering/Data, Director of AI/ML, AI Engineer, Machine Learning Engineer, AI Architect

Related on AIssential

• Qwen3.5-397B-A17B: the smallest Open-Opus class, very efficient model — Qwen3.5 offers multimodal, long-context capabilities with efficient sparse MoE architecture, enabling local deployment despite its large scale.
• The Sequence AI of the Week #813: Deep Diving Into the Amazing GLM-5 — AI is shifting to autonomous "agentic engineering" requiring advanced reasoning and context handling.
• Qwen 3.5 Local Test with Ollama | Coding, OCR, Data Extraction, Image Understanding — Qwen 3.5 MoE model offers strong multimodal capabilities and efficient local inference.
• Qwen 3.5 The GREATEST Opensource AI Model That Beats Opus 4.5 and Gemini 3? (Fully Tested) — Qwen 3.5 is a powerful open-source, multimodal AI model excelling in speed and diverse applications.
• MoE vs Dense Models for Structured Data Extraction — Who Wins? — Dense LLMs are superior to MoE models for extracting large volumes of structured data from tables.
• Alibaba's Qwen 3.5 397B-A17 beats its larger trillion-parameter model — at a fraction of the cost — Sparse Mixture-of-Experts models can surpass larger predecessors in performance and efficiency.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by TheSequence.