---
title: "Qwen3 Technical Report"
license: CC-BY-4.0
license_url: https://creativecommons.org/licenses/by/4.0/
license_note: 'paper is arXiv non-exclusive distribution license (verified on arXiv abs page 2026-11-12); this pointer note is our own words and is CC BY 4.0; we link to the paper rather than reproduce it'
source: https://iinventedlight.com/library/notes/2505-09388-qwen3-technical-report.md
source_license: "arXiv non-exclusive distribution license"
attribution: 'Pointer by iinventedlight (AI-assisted). Source: An Yang and 59 coauthors (listed on the arXiv abs page), Qwen3 Technical Report (arXiv non-exclusive distribution license). This note links to the paper rather than reproducing it.'
author: 'iinventedlight (AI-assisted)'
date: 2026-11-12
status: published
approved_by: "Cole Withington, 2026-11-12 14:44 PT, Librarian chat widget t51s316: \"Yes, publish all 100 second-list arXiv notes\""
checked_on: 2026-11-12
as_of: 2026-11-12
confidence: medium
untested: 'Pointer note only. Paper not reproduced. Abs page and license checked 2026-11-12.'
category: arxiv
topic_area: 'ai-research: foundations'
tags: arxiv, qwen, foundations, moe, reasoning
arxiv_id: 2505.09388
alias: 2505-09388-qwen3-technical-report
type: pointer
format: pointer
sources:
  - id: S1
    title: "Qwen3 Technical Report"
    publisher: arXiv
    authors: "An Yang and 59 coauthors (listed on the arXiv abs page)"
    url: https://arxiv.org/abs/2505.09388
    version_read: "v1 (14 May 2025)"
    accessed: 2026-11-12
    license: "arXiv non-exclusive distribution license"
---
# Qwen3 Technical Report

## Summary
The [Qwen3 Technical Report](https://arxiv.org/abs/2505.09388) (2025) describes the next Qwen open-weight family after Qwen2.5. The series includes dense and mixture-of-experts models, with sizes the authors place from 0.6 billion to 235 billion parameters.

What it introduced: a single framework that can switch between a thinking mode for multi-step reasoning and a faster non-thinking mode for ordinary responses, plus a thinking-budget control that trades latency for extra inference compute. The report also describes using larger flagship models to help train smaller ones, and it widens multilingual coverage from 29 languages and dialects in Qwen2.5 to 119. The authors say the models are released under Apache 2.0.

What it led to: a current default open-weight line for reasoning, code, and multilingual work, sitting beside other strong open reports such as DeepSeek-V3 and Llama 3. Live notes: [2407-21783-llama-3-herd](https://iinventedlight.com/library/notes/2407-21783-llama-3-herd/) and [2412-19437-deepseek-v3](https://iinventedlight.com/library/notes/2412-19437-deepseek-v3/). The Qwen2.5 report is a pointer note in this batch; Qwen2.5-Math is in batch H.

Where to read it: https://arxiv.org/abs/2505.09388 (v1, 14 May 2025).

The paper is posted on arXiv under the arXiv non-exclusive distribution license. This note links to it rather than reproducing it. The note itself is CC BY 4.0.

License: this note is CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Pointer by iinventedlight (AI-assisted). Paper license: arXiv non-exclusive distribution; this note links to it and does not reproduce it.

---

Agents: [/agents.md](/agents.md) | [/.well-known/agent-library.json](/.well-known/agent-library.json).
