#Alibaba

'Alibaba released Qwen3-Max, a 1T+ Mixture-of-Experts model with a production-ready thinking mode and notable day-one wins on coding and agentic benchmarks. Model Studio and Qwen Chat provide immediate access, with thinking runs requiring streaming output.'

Alibaba's Qwen team published FP8-quantized checkpoints for Qwen3-Next-80B-A3B (Instruct and Thinking), enabling more efficient long-context inference and practical deployment of an 80B/3B-active hybrid MoE on commodity GPUs

'Alibaba open-sourced Tongyi DeepResearch-30B-A3B, a 30B-parameter MoE LLM tailored for long-horizon, tool-augmented web research with 128K context and strong benchmark scores.'

'Qwen3-ASR Flash from Alibaba is a single-model ASR that auto-detects and transcribes 11 languages, supports context injection for domain terms, and keeps WER below 8% in noisy or musical audio.'

'Alibaba's Qwen-Image-Edit extends Qwen-Image to provide precise semantic and appearance edits, bilingual text handling and easy deployment via Hugging Face.'

Wix and Alibaba have partnered to empower small and medium businesses with AI-driven tools and global sourcing capabilities, simplifying online selling and enabling SMBs to compete globally.

Alibaba introduces Qwen3-MT, a next-generation multilingual machine translation model featuring cutting-edge architecture and reinforcement learning for high-quality, cost-efficient translations across 92+ languages.

Alibaba's Qwen team introduces Qwen-VLo, an advanced multimodal model enabling creative text-to-image generation and editing with multilingual support and progressive scene construction.

Alibaba's Qwen Team has released the Qwen3-Embedding and Qwen3-Reranker series, offering state-of-the-art open-source multilingual embedding and ranking models that outperform existing solutions.

Manus has ignited a surge in AI agent development in China, with startups and tech giants racing to create autonomous digital assistants that integrate deeply into daily life and global markets.

'Alibaba’s ZeroSearch framework leverages reinforcement learning and simulated document generation to train language models for retrieval without relying on costly real-time search APIs, achieving performance comparable to or better than Google Search.'

Alibaba introduces Qwen2.5-Omni-3B, a 3-billion parameter multimodal AI model that reduces VRAM usage by over 50% while delivering near 7B model performance, designed for consumer-grade GPUs.

Alibaba launches Qwen3, an innovative open-source AI series blending fast and deliberate reasoning, challenging ChatGPT and Google’s AI supremacy.

Alibaba's Qwen3 introduces a new generation of large language models that excel in hybrid reasoning, multilingual understanding, and efficient scalability, setting new standards in AI performance.