Qwen3 Blog

Mar 19, 2026
Qwen 3.5

How to Use Qwen 3.5 API

On Lunar New Year’s Eve (Feb 16–17, 2026), Alibaba Group released its next-generation model, Qwen 3.5 — a multimodal, agent-capable model positioned for what the company calls an “agentic AI” era. Industry coverage highlighted claims of large gains in efficiency and cost, and rapid support from hardware and cloud vendors. CometAPI is options for developers who want hosted API access or an OpenAI-compatible integration, while AMD announced Day-0 GPU support for the model on its Instinct line. ByteDance is one of the principal domestic competitors that released upgrades around the same holiday window. OpenAI remains a reference point for comparison in benchmarks and integration style.
Qwen-3.5 on Lunar New Year — does it beat the closed-source top tier in 2026
Mar 19, 2026
Qwen 3.5

Qwen-3.5 on Lunar New Year — does it beat the closed-source top tier in 2026

Alibaba’s new Qwen3.5 is a major step forward — it closes the gap with, and in some agentic / multimodal workloads claims parity or advantage over, certain frontier closed-source models on a number of public benchmarks and internal tests. However, “outperform” depends on the workload: on agentic tool-use, multimodal document/video understanding, and cost-per-inference Qwen3.5 is reported to be extremely competitive (and in some vendor charts ahead). The practical takeaway: Qwen3.5 appears to be a genuine frontier contender in early 2026 — for many enterprise agentic and multimodal use cases it is now viable as a primary option.
How to use Qwen3-max thinking
Feb 3, 2026
qwen3-max

How to use Qwen3-max thinking

Alibaba’s Qwen3-Max-Thinking — the “thinking” variant of the massive Qwen3 family — has become one of the headline stories in AI this year: a trillion-plus parameter flagship tuned for deep reasoning, long-context understanding and agentic workflows. In short, it’s the vendor’s move to give applications a slower, more traceable “System-2” mode of thought: the model doesn’t just answer, it can show (and use) steps, tools, and intermediate checks in a controlled way.
Alibaba AI Unveils Qwen3-Max: trillion-parameter Qwen model, CometAPI supports
Jan 6, 2026
qwen3-max

Alibaba AI Unveils Qwen3-Max: trillion-parameter Qwen model, CometAPI supports

Alibaba’s Qwen team has released Qwen3-Max-Preview (Instruct) — the company’s largest model to date, with more than 1 trillion parameters — and made it
Qwen3-Max-Preview API
Jan 6, 2026
qwen3-max

Qwen3-Max-Preview API

Qwen3-Max-Preview is Alibaba’s latest flagship preview model in the Qwen3 family — a trillion+-parameter, Mixture-of-Experts (MoE) style model with an ultra-long 262k token context window, released in preview for enterprise/cloud use. It targets *deep reasoning, long-document understanding, coding, and agentic workflows.
Qwen3-Coder: Performance, Architecture & Access
Jan 6, 2026
Qwen3-Coder

Qwen3-Coder: Performance, Architecture & Access

Alibaba’s latest advance in artificial intelligence, Qwen3-Coder, marks a significant milestone in the rapidly evolving landscape of AI-driven software
Qwen3-Coder API
Jan 6, 2026
Qwen3-Coder

Qwen3-Coder API

The Qwen3-Coder API is a code generation and completion API based on Alibaba’s Qwen3 language model family, optimized for software development tasks such as writing, understanding, and debugging code across multiple programming languages.
Alibaba Releases Qwen3‑Coder and Qwen Code: A Breakthrough in Agentic AI Coding
Jan 6, 2026
Qwen Code
Qwen3-Coder

Alibaba Releases Qwen3‑Coder and Qwen Code: A Breakthrough in Agentic AI Coding

On July 23, 2025, Alibaba Group officially launched Qwen3‑Coder, an open‑source artificial intelligence model tailored for software development and autonomous