alibaba

阿里巴巴发布了通义千问 3.7 Max，一款专为智能体时代设计的旗舰编码模型。该模型在长周期自主执行、前端生成和3D场景构建上表现突出，多项基准测试中与顶尖闭源模型持平甚至超越，是接近前沿的中国模型。

Alibaba's Qwen team has released Qwen3.7-Max, a proprietary agent-foundation model achieving top scores on multiple benchmarks including Terminal-Bench 2.0, SWE-Pro, and GPQA Diamond, with consistent performance across various code environments.

A tweet by Sentdex highlights Alibaba Qwen's transparent benchmark reporting for the Qwen3.7-Max model, contrasting it with others who cherry-pick benchmarks.

Alibaba unveiled the Zhenwu M890 AI chip, designed to handle the memory and communication demands of AI agent workloads, as part of its push for domestic alternatives.

Alibaba Qwen announces two major model releases: Qwen3-Omni, the first natively end-to-end omni-modal AI unifying text, image, audio and video, and Qwen3-Next-80B-A3B, an ultra-efficient MoE model with 3B activated parameters per token, achieving SOTA performance and 10x faster inference than Qwen3-32B.

Alibaba has released Qwen3.7 Preview models on Arena, including Qwen3.7-Max-Preview and Qwen3.7-Plus-Preview, with the full Qwen3.7 series expected soon.

Alibaba's Qwen team announces the upcoming release of version 3.7 models.

Alibaba releases Qwen3.7-Max-Preview and Qwen3.7-Plus-Preview on Arena, achieving top rankings in Text and Vision categories.

A new version of Qwen, Qwen 3.7, has been spotted on the official Qwen website, suggesting an upcoming release.

Qwen 3.7 has been released on Qwen Chat, marking an update to Alibaba's AI model series.

This technical report presents Qwen-Image-2.0, a new image generation model from Alibaba's Qwen team, detailing its architecture and capabilities.

Alibaba's AIDC team has released Ovis2.6-80B-A3B, an Apache 2.0 licensed Mixture of Experts multimodal LLM featuring 80B total parameters with 3B active, 64K context length, and native support for 2880×2880 images with Chain-of-Thought visual reasoning.

Alibaba's Qwen 3.6 Plus model is temporarily available for free via the Nous Portal subscription platform, which aggregates access to hundreds of AI models.

Alibaba released insights on organizational R&D in the AI Native era, pointing out that traditional organizational structures need to shift from accommodating human limitations to adapting to the efficient execution of AI Agents. The article emphasizes that the core bottleneck of AI transformation lies in outdated information formats; implicit experience must be transformed into AI-understandable infrastructure, while preserving the human role in innovation and cultural building.

DeepSeek has rejected investment from Alibaba to maintain corporate independence and avoid restrictive ecosystem integration clauses, despite significant interest from Chinese tech giants. The company seeks funding primarily for computational power and talent retention while minimizing external control over its commercialization agenda.

The article discusses that Qwen, Alibaba's large language model, is not available for free usage, addressing pricing or access limitations for the model.

Tencent and Alibaba are reportedly in talks to invest in Chinese AI startup DeepSeek at a valuation exceeding $20 billion.

Alibaba releases Qwen3.6-Max-Preview, an updated version of its Qwen3.6 model series with improved performance and capabilities.

Page Agent is an open-source JavaScript library from Alibaba that lets you control web interfaces with natural language, without requiring browser extensions or headless browsers.

Alibaba releases Zvec v0.5.0, an open-source in-process vector database with new features including full-text search, hybrid retrieval, DiskANN index, and new SDKs for Go, Rust, plus a visual tool.