# Alibaba launches open source Qwen3 model that surpasses OpenAI o1 and DeepSeek R1

_Sunday, July 12, 2026 at 8:00 AM EDT · AI · Latest · Tier 2 — Notable_

![Alibaba launches open source Qwen3 model that surpasses OpenAI o1 and DeepSeek R1 — Primary](https://images.ctfassets.net/jdtwqhzvc2n1/DtssLzpDm1krEaaCUr8kc/b66b84271f6aebade2865aae57f7c656/029ebd88-b24c-4396-80ff-2a47b3477cc6.png?w=800&q=75)

Alibaba's Qwen team has officially launched Qwen3, a new series of eight open source large language multimodal models that the company says approach the performance of proprietary models from OpenAI and Google. The team announced the release includes two mixture-of-experts models and six dense models ranging from 0.6 billion to 32 billion parameters. According to the team, the 235-billion parameter mixture-of-experts version, codenamed A22B, outperforms DeepSeek's open source R1 and OpenAI's proprietary o1 on key third-party benchmarks including ArenaHard and nears the performance of Google's proprietary Gemini 2.5-Pro. The models feature hybrid reasoning capabilities allowing users to toggle between fast responses and more compute-intensive reasoning steps for difficult queries in science, math, and engineering. Users can access the models via Hugging Face, ModelScope, Kaggle, GitHub, and the Qwen Chat web interface and mobile applications. All models are available under the Apache 2.0 open-source license. The pretraining dataset doubled to approximately 36 trillion tokens compared to the predecessor Qwen2.5. Junyang Lin, a member of the Qwen team, commented on X that building Qwen3 involved addressing challenges such as scaling reinforcement learning stably and expanding multilingual performance without quality sacrifice. The models support 119 languages and dialects.

## Sources

- [VentureBeat](https://venturebeat.com/ai/alibaba-launches-open-source-qwen3-model-that-surpasses-openai-o1-and-deepseek-r1)

---
Canonical: https://techandbusiness.org/newswire/PGyO8GOXYxu6ZJgc2n8YmG
Retrieved: 2026-07-12T11:11:50.765Z
Publisher: Tech & Business (techandbusiness.org)