Qwen3 235B A22B Thinking 2507

Qwen3.8-Max-0902 (also known as qwen3.8-max-2026-09-02) is a snapshot version of Alibaba Cloud Tongyi Qianwen's qwen3.8-max. It pushes encoding depth further, enabling it to handle more complex engineering-level projects and long-term autonomous development; collaborative agent capabilities are significantly enhanced, performing more confidently in multi-tool orchestration and end-to-end delivery; visual understanding is comprehensively improved, with more sensitive and accurate chart reasoning, document parsing, and multimodal perception. Continuing the 1-million-context window, reasoning modes, and a complete tool ecosystem, it continues to evolve at a higher level of intelligence.

Qwen3.8 Flash is Alibaba Cloud Qwen’s flagship native vision-language model for coding, office tasks, long-context reasoning, and agent workflows. It supports a 1M-token context, 128K output, web access, and tool calling. Compared with Qwen3.7-Plus, Qwen3.8-Flash significantly reduces training and inference costs—the training overhead is only about one-ninth of the former—while offering stronger capabilities on coding and office tasks.

Wan 3.0 (Tongyi Wanxiang 3.0) is an integrated video generation and editing model launched by Alibaba’s Tongyi Lab. It is suitable for scenarios such as advertising, e-commerce, short films, character animation, video editing, and document-to-video. Its advantages include native support for videos up to 30 seconds long and the unified handling of reference content such as text, images, audio, video, documents, spreadsheets, presentations, and web pages, delivering more realistic visuals and audio, stable character consistency, and capabilities for referencing, editing, replication, and driving. Compared with Wan 2.7, which split text-to-video, image-to-video, reference generation, and editing across multiple models, Wan 3.0 integrates these capabilities into a single model, making the creative workflow more unified and flexible.

Wan3.0 Video Prime is Alibaba Cloud’s preview high-speed edition of its All-in-One video model, positioned for faster-turnaround ads, ecommerce, short films, and editing. Aligned with Wan3.0 Video’s capabilities, it supports text-to-video, image-to-video, reference-based generation, document/webpage-to-video, up to 30 seconds at 30fps, 1080P output, adaptive aspect ratios, and native audio. It unifies modes that Wan 2.7 handled through separate task-specific models.

Qwen3.8-Max is Alibaba Cloud Tongyi Qianwen's next-generation flagship large language model, featuring a mixture-of-experts (MoE) architecture with 2.4 trillion parameters. It achieves another breakthrough in encoding depth, enabling it to handle more complex engineering-grade projects and long-term autonomous development; collaborative agent capabilities are significantly enhanced, performing more confidently in multi-tool orchestration and end-to-end delivery; visual understanding is comprehensively improved, with more sensitive and accurate chart reasoning, document parsing, and multimodal perception. Continuing the 1-million-context window, reasoning modes, and a complete tool ecosystem, it continues to evolve at a higher level of intelligence.

Qwen3.8-2.4T-A95B is Alibaba’s most powerful Qwen model to date. It is a 2.4‑trillion‑parameter sparse Mixture-of-Experts (MoE) model with approximately 95 billion active parameters. It is built for autonomous, long‑duration tasks: multi‑day code runs, reproducing research papers, and self‑improvement.