Last week, Alibaba showcased advancements in innovation across multiple sectors, spanning corporate culture, aviation safety, and AI development. Highlights include its recognition as a top workplace for innovators by Fast Company, a strategic cloud partnership with AirNav Indonesia to modernize aviation navigation, and the launch of its ultra-efficient Qwen3-Next AI architecture alongside Qwen3-ASR-Flash and trillion-parameter Qwen3-Max preview.
Alibaba has been recognized as No.18 in on Fast Company’s 2025 list of the 100 Best Workplaces for Innovators as well as a spot on the AI, Automation, and Machine Learning Excellence list. “Trade innovations, such as a new B2B AI search engine called Accio, help Alibaba Group move toward its mission to do business anywhere”, said Fast Company in the honorees feature.
The recognition affirms Alibaba’s group-wide strategy to harness AI and empower employee-driven innovation, reinforcing its leadership in creating a workplace culture that drives innovation. Alibaba enables employee-driven innovation to create real business impact. Accio, Alibaba.com’s AI-powered B2B search engine, a leading AI application, was initiated by a three-person team. The recognition comes as Alibaba accelerates its AI investment across the organization. Over the next three years, Alibaba has pledged to invest $53 billion in AI infrastructure, foundation models, and AI-native applications.
AirNav Indonesia, a national aviation navigation service provider in Indonesia, has partnered with Alibaba Cloud to strengthen its digital resilience and advance the digital transformation of the aviation navigation industry in the country.
Through this collaboration, AirNav Indonesia will use Alibaba’s Cloud Disaster Recovery Center (DRC) and Alibaba Cloud Security Service Service to strengthen its digital infrastructure and cybersecurity capabilities.These robust security measureswillallow AirNav Indonesia to maintain operational continuity, providing proactive protection of web applications and flight navigation systems, ensuring service stability that is vital to national and international aviation safety.
The partnership reinforces AirNav Indonesia’s commitment to providing safe, reliable, and efficient flight navigation services, making the company a pioneer of digital transformation in the field of flight navigation in the region.
Alibaba has introduced Qwen3-Next, a new model architecture optimized for long-context understanding, scalability, and computational efficiency. By integrating innovations like a hybrid attention mechanism and Ultra-Sparce Mixture of Expert design and Multi-Token Prediction, Qwen3-Next achieves exceptional performance with minimal resource use.
The first model in this series, Qwen3-Next-80B-A3B-Base, employs 80 billion parameters but activates only 3 billion during inference. It surpasses the dense Qwen3-32B model while slashing training costs to under 10% of Qwen3-32B’s GPU hours, delivering over 10x higher throughput for contexts exceeding 32K tokens. The Qwen3-Next-80B-A3B-Instruct model natively supports a 256K-token window (extendable to 1 million) and matches the performance of Alibaba’s flagship Qwen3-235B-A22B-Instruct-2507. Both Instruct and Thinking modes are open-sourced on Hugging Face, Kaggle, and ModelScope. The Thinking variant Qwen3-Next-80B-A3B-Thinking model excels in complex reasoning, outperforming leading closed-source models and approaching the performance of the flagship Qwen3-235B-A22B-Thinking-2507.
Additionally, Alibaba recently launched Qwen3-ASR-Flash, a multilingual automatic speech recognition model powered by Qwen3-Omni and trained on tens of millions of hours of high-quality multilingual speech data. It excels in noisy environments and supports 11 major languages, Chinese dialects, and English regional accents. Alibaba also previewed Qwen3-Max, its largest non-thinking model with over 1 trillion parameters. Ranked No.6 in Text Arena, Qwen3-Max significantly reduces hallucinations, enhances multilingual reasoning, and it optimized for advanced workflows like Retrieval-Augmented Generation, and tool-calling.
This article was originally published on Alizila written by Jing Wang , Crystal Liu and Karen Zhang
Qwen3-Next: A New Generation of Ultra-Efficient Model Architecture Unveiled
1,291 posts | 455 followers
FollowAlibaba Cloud Community - October 27, 2021
Alibaba Cloud Community - October 21, 2021
Alibaba Cloud Community - March 27, 2025
Alibaba Cloud Community - May 21, 2025
Alibaba Cloud Community - January 24, 2024
Alibaba Clouder - June 10, 2020
1,291 posts | 455 followers
Follow
Tongyi Qianwen (Qwen)
Top-performance foundation models from Alibaba Cloud
Learn More
Alibaba Cloud for Generative AI
Accelerate innovation with generative AI to create new business success
Learn More
AI Acceleration Solution
Accelerate AI-driven business and AI model training and inference with Alibaba Cloud GPU technology
Learn More
Platform For AI
A platform that provides enterprise-level data modeling services based on machine learning algorithms to quickly meet your needs for data-driven operations.
Learn MoreMore Posts by Alibaba Cloud Community
