---
id: "qwen3-technical-report"
title: "Qwen3 Technical Report"
creator: "Qwen Team / Alibaba"
creator_url: "https://arxiv.org/abs/2505.09388v1"
creator_x: null
url: "https://arxiv.org/abs/2505.09388v1"
level: "advanced"
level_max: null
category: "frontier"
subcategory: "model-releases"
region: "CN"
license: "closed"
license_scope: "linked-resource"
access: "free"
models: ["qwen3"]
source_date: "2025-05-14"
date_added: "2026-10-18"
---

[Frontier](/frontier/index.md)

# Qwen3 Technical Report

by [Qwen Team / Alibaba](https://arxiv.org/abs/2505.09388v1)

[Read the original →](https://arxiv.org/abs/2505.09388v1)

Read the model team’s account of Qwen3’s dense and MoE releases and unified thinking and non-thinking modes.

The Qwen team introduces a model family spanning dense and mixture-of-experts architectures. The report describes unified thinking and non-thinking modes, inference budgeting and multilingual evaluation.

Level: [Advanced](/levels/index.md#advanced)
Models: qwen3
Category: [Frontier](/frontier/index.md)
Tags: [primary-source](/tags/primary-source/index.md), [regional-coverage](/tags/regional-coverage/index.md)
Source date: 2025-05-14
Added: 2026-10-18
Last verified: 2026-10-18
Status: ok
Prerequisites: Assumes familiarity with model architectures, inference trade-offs and benchmark interpretation.
Subcategory: [model releases](/frontier/model-releases/index.md)
Region: CN
Original language: en
Access: free
Resource reuse: closed: No general open-reuse grant established for the linked resource. Closed is a conservative reuse bucket, not a claim that access is restricted; model licensing does not license a paper. [Facet definitions](/about/index.md#facets)

Summary in our own words. Rights remain with the original creator. [Credit and correction policy](/about/index.md#credit-policy).

[Plain-text Markdown copy](/frontier/model-releases/qwen3-technical-report/index.md)