Ali

On September 18th, Ali thousands asked about the new generation of Qwen3.8-Omni-Flash, which is officially on the line today, and has already been made available on the Al-Ai platform. The model allows for the simultaneous processing of text, images, audio and video input to support the 1M context. Full-state models refer to multiple input forms that can be processed simultaneously, such as text, images, audio, video, etc. Officially, while maintaining the capability of the same size text model, the total modulation capacity was significantly higher than that of the previous generation. In the cumulative 30 assessments, Qwen 3.8-Omni-Flash compares..

On September 2, QuestMobile published a study on the development of the 2026 AI platform, which shows that in July of this year the total number of live users of AI ' s efficient office tracks amounted to 120 million, an increase of 8.2% over the same period; the total monthly usage was approximately 13.87 billion, an increase of 112.4% over the same period. User size is still growing steadily and frequency of use has doubled. The portal is also moving from a web chat box to a client closer to the workstream: AI Worker App is $7.926 million a month, which is an increase of 261.1%; PC..

On May 16, Ali officially released Qoder 1.0 yesterday, to be upgraded to an "Intelligence Self-Developing Workstation" , the user simply needs to complete the definition of demand, and the whole process of execution, validation and delivery is done automatically by the AI Smart. Windows, MacOS and Linux users are available for download. Quest Upgrade to a stand-alone view window that integrates task management, status tracking, product review and knowledge callable, running in parallel with the Editor window; multitasks extended in parallel to cross-project, cross-code library dimensions, and a unified panel tracks all tasks in real time..

On March 10, according to Read Jia today, the Ali version of OpenClaw, which was measured earlier in the year, changed its name to JVS Claw in recent days and may also be preparing for the end-line of the product. According to the report, “JVS Claw” currently has a stand-alone App and web version, but both are in the internal phase, requiring an invitation experience, during which each user can create one bot with a free model of 8000 Credits, valid for 14 days. This AI product is customized and evolutable C..

In the early hours of yesterday, Ali thousands asked Lin Jun-chun, the tech head of the Big Model, to suddenly announce his departure from the social platform. This was followed by a heat wave on the Internet. According to the information received, Lin Jun-hyun is the core driver of the Ali AI open source model and one of the youngest P10s in Ali. According to the "Intelligence Emerging" report, yesterday afternoon, the generalist laboratory held an emergency All Hands conference, and the Ali Group CEO Wu Mi-O told thousands of employees, "I should have known this earlier." The meeting was attended by several senior managers from Ali, the Qwen team and other team members from the common sense laboratory. I don't know

On March 4th, Ali announced that QoderWorker desktop Agent was fully open and suitable for the Mac/Windows platform, which, without additional deployment, would have a “special AI assistant who can do business”, called “available to all”. It was described that the QoderWork application had internalized top-of-the-art models and the Agent framework, allowing for an autonomous understanding of complex mission intentions, automatic dismantling of implementation steps and real-time feedback of mission progress to users. And this new one has added a model grader, and the user can choose from the use..

On February 17, Ali, on the Chat.qwen.ai page, went low on line with two new models, Qwen3.5-Plus and Qwen3.5-397B-A17B. 1AI learned from the official page that Qwen3.5 Plus is located in the latest large language model of the Qwen 3.5 series, and Qwen 3.5-397B-A17B is located in the Qwen3.5 large language model of the Open Source series. Both models support text and multi-mode tasks. According to Ali Yun, Qwen3..

On January 27th, Ali published a model of the flag ship ' s reasoning Qwen3-Max-Thinking. According to official presentations, it has achieved significant improvements in several critical dimensions, including knowledge of facts, complex reasoning, compliance with directives, human preferences and intelligence capabilities. In 19 authoritative benchmark tests, performance is comparable to top model GPT-5.2-Thinking, Claude-Opus-4.5 and Gemini 3 Pro. The new model's total parameters are in excess of hundreds of millions of dollars

ON DECEMBER 16TH, ALI RELEASED A NEW GENERATION OF MIRROR 2.6 SERIES OF MODELS, WHICH WAS COMPLETELY UPGRADED TO PROFESSIONAL VIDEO PRODUCTION AND IMAGE CREATION SCENES, THE FIRST VIDEO MODEL IN THE COUNTRY TO SUPPORT ROLE-PLAYING. THE MODEL ALSO SUPPORTS THE FUNCTIONS OF SYNCHRONIZATION, MULTIPLE LENS GENERATION AND SOUND DRIVE, AND IS THE MOST FUNCTIONAL VIDEO GENERATION MODEL IN THE WORLD. 2.6 AL-AIYUN YING, A NETWORK OF OFFICIALS. 1AI WITH THE FOLLOWING OFFICIAL INTRODUCTION: 2.6 VIDEO GENERATION TO UPGRADE BRIGHT SPOTS, ROLE-PLAYING, EVERYTHING TO ENTER, 2.6..

On 1 December, according to the news, Ali Tunyi announced that his latest biographic model, Z-Image, was officially on board. Officially, Z-Image is an open-source, free-of-charge, efficient image generation base model with a parameter of 6B. Models can generate super-real images of quality comparable to top-level business models on the 16GB visible consumer-grade graphic cards, particularly with regard to bilingual Chinese and English text rendering. At the same time, Z-Image has a wide appreciation of the real world and is able to produce well-known landmarks (e.g. Eiffel Tower, Palace), well-known personalities and particular cultural dollars..

On November 29, the news that yesterday afternoon, the official public issue of the Ali Tunyi model stated that the newly released biographic model Z-Image had quickly reached the top of the Hugging Face Trend, with a first-day download of 500,000. Officially published articles state that Z-Image achieves a photo-level realism comparable to a magnitude model of 6B parameters. Whether it is skin sense, filament details, or natural light and material texture, it is finely reduced, and the image and atmosphere are both aesthetically expressed. Besides, Z..

On November 27, according to the news, Ali's cash self-study of AI glasses - Quake Ai glasses S1 - was officially published today at an official subsidized price of $3999, and today's pre-sale: standard set of $4299, official subsidized price of $3999, super re-suspended suit of $4599, official subsidized price of $4299, which uses a light-quantified design with a minimum width of 7.5 mm for the mirrors, a minimum of 3.3 mm for the frame frame, and a closer to the normal eyeglass. In detail, the glasses use the hidden rims of thorns, titanium alloy, double-splendour

On 25 September, according to the news, at the 2025 Cloud Habitat Congress, the Director of the Ali Group and CEO Wu Yui Song indicated that the 38 billion AI infrastructure was being actively advanced and that additional inputs were planned. At the conference, Ali launched six new models + 1 brands: Qwen MAX: Large Model for trillion Parameters, Coding and Tool Call Capability to Top the International List; Qwen3-Omni: Large Model for a New Generation of Numerical Whole Models, which truly achieves the "Qwen3-VL:Agent and Coding"

September 7 news, Ali quietly in Tongyi Qianqian official website, OpenRouter online Qwen3-Max-Preview model, and called it the most powerful language model in the Tongyi Qianqian series. Afterwards, Tongyi big model through the official microblogging introduced the features that this model has. Compared with the 2.5 series, the released Qwen3-Max-Preview (Instruct) realizes significant enhancements in the dimensions of Chinese and English comprehension, complex instruction following, and tool invocation, and at the same time significantly reduces the knowledge illusion, making the model smarter and more available to...

July 13 news, July 9, local time, held in Geneva, Switzerland, the United Nations AI for Good Summit, Ali Dharma Institute two AI technology was selected as AI for Good innovation and influence case set and won the excellent case award: "eight view" meteorological model, "flat scan", "CT + AI" multi-cancer early screening, "CT + AI", "CT + AI". The "Eight Observations" weather model and "CT+AI" multi-cancer early screening. Ali Dharmo Academy also cooperated with the International Telecommunication Union (ITU) under the United Nations to build an open platform for AI experts to communicate and share knowledge. Peng Bingqing, a senior algorithm expert at the Dharma Institute's Decision Intelligence Laboratory, shared at the meeting that "eight...

According to Blue Whale Technology, Manus AI is rumored to be moving again this week, with several employees confirming that the company will move its headquarters to Singapore and make large-scale adjustments to its domestic team. The former Manus employees affected by this adjustment revealed that according to their understanding at the time, Manus may no longer set up a domestic office, and no longer retain the domestic business team, all relocated to the new headquarters in Singapore. He said that most of the employees in China were not aware of the change in the headquarter relocation, and "probably just knew that there was an office in Singapore. It was not until Monday that the company formally notified its employees and held a conversation, giving...

July 7 news, today AliCloud announced that Tongyi officially open source network intelligent body WebSailor, the intelligent body has a strong reasoning and retrieval capabilities, after the release of the intelligent body review set BrowseComp topped the list of open source network intelligent body. 1AI notes that the current WebSailor construction program and some of the data set has been open source in Github. According to AliCloud, WebSailor web intelligences can be applied to retrieval tasks in complex scenarios, and for fuzzy questions can be quickly searched in different webpages and reasoned to check...

July 4 news, Ali "Tongyi big model" public number today announced that the Tongyi laboratory's first audio generation model ThinkSound is now officially open source, will break the imagination of the "silent picture" limitations. ThinkSound applies CoT (Chain-of-Thought) to audio generation for the first time, allowing AI to learn to "think through" the relationship between picture events and sound step by step, thus realizing high-fidelity, strongly synchronized spatial audio generation - not just "picture dubbing". - This enables high-fidelity, strongly synchronized spatial audio generation - not just "looking at the picture and dubbing", but actually "understanding the picture". ...

June 6 news, Ali early this morning open source Qwen3-Embedding series of models (Embedding and Reranker), designed for text representation, retrieval and sorting tasks, based on the Qwen3 base model for training. According to officials, the Qwen3-Embedding family has demonstrated excellent performance in text representation and sorting tasks in multiple benchmark tests. The Qwen3-Embedding family has the following characteristics: Excellent generalization: The Qwen3-Embedding family achieves industry...

April 29 news, early this morning, Alibaba released a new generation of Tongyi thousand questions Qwen3 model, a top of the world's strongest open source model. Models Layers Heads (Q / KV) Tie Embedding Context Length Qwen3-0.6B 28 16 / 8 Yes 32K Qwen3-1.7B 28 16 / 8 Yes 32K Qwen3-4B 36 32 / 8 Yes 32K ...

On April 28th, Ali Tongyi Qianqian open source head Yang Lin Junyan hinted in a post on X that Qwen3 models are expected to be released today. It's worth noting that the Qwen3 model collection was briefly unveiled at ModelScope, Alibaba's open source AI modeling community, and has since been taken offline, including Qwen3-4B-Base, Qwen3-1.7B, Qwen3-0.6B, and Qwen3-30B-A3B-Base. All models are licensed under the Apache License ...

Ali Tongyi Wanphase "first and last frame video model" announced on April 17th open source, the model parameter number of 14B, said to be the industry's first tens of billions of parameter scale of the open source first and last frame video model. It can generate a 720p HD video that connects the first and last frames according to the user-specified start and end images, and this upgrade will meet the user's needs for more controllable and customized video generation. Users can experience the model for free on TongyiWanPhase's website, or download the model from Github, Hugging Face, or Magic Match community for local deployment and secondary development. Technology...

March 28th news, early this morning, Ali Tongyi Qianqi team announced the launch of a new generation of visual reasoning model QVQ-Max. According to the official introduction, QVQ-Max is not only able to understand the picture and video content, but also can provide analysis and reasoning for the above information. In addition to analyzing and reasoning, QVQ-Max can also design illustrations, generate short video scripts, and even create role-playing content according to users' needs. Core Capabilities: From Observation to Reasoning The capabilities of QVQ-Max can be summarized in three areas: detailed observation, in-depth reasoning and flexible application. Here are the respective...

On February 20, Alibaba Group CEO Wu Yongming said at the Q3 FY2025 earnings meeting that Ali will soon release a deep inference model based on Thousand Questions Qwen 2.5-MAX, according to Sina Technology. 1AI notes that on New Year's Day, January 29, AliCloud announced its new Tongyi Thousand Questions Qwen 2.5-Max hyperscale MoE model, which it claimed to have outperformed competitors such as DeepSeek V3 in several benchmarks. AliCloud then went on to announce that Qwen 2.5-Max ...