← all stories

Alibaba

Alibaba is a major AI model developer and cloud provider, with its Qwen line spanning text, audio, video, robotics, and world models, and it is now positioned to buy NVIDIA RTX Pro 5500 GPUs under a Chinese government approval process. Its most recent releases include Qwen3.8-Omni-Flash and Qwen3.8-Max, and it backs the AI Film Festival Japan 2026 through Alibaba Cloud.

Synthesized from 11 Yomimono stories · updated Oct 10

Alibaba's Qwen team has released a series of models through 2026. In June it announced the Qwen-Robot Suite, three models for robot navigation, manipulation, and world recognition. That was followed by HappyOyster 1.0, a service that generates playable worlds in real time from text or image prompts, free with daily credits through July 17, 2026. In August the company unveiled Qwen3.8-Max, a multimodal model with 2.4 trillion total parameters and 95 billion active parameters, and Qwen3.8 27B, an open model that third-party analysis placed level with GPT-5.6 Luna on the Intelligence Index and ahead of GPT-5.6 Terra on the Agentic Index. Qwen-Audio-3.0-TTS, a text-to-speech model, ranked first on Artificial Analysis's TTS leaderboard.

The open-weight strategy has a commercial condition. Alibaba plans to require major commercial users of Qwen3.8-Max to share revenue, following Moonshot AI's Kimi K3 license, which mandates contracts for businesses with annual sales over $20 million. The revenue sharing rate is undecided. On the infrastructure side, Bloomberg reported that Moonshot AI signed a contract with Alibaba to use approximately 20,000 NVIDIA chips for its Kimi series, reportedly older Hopper architecture, possibly H200, though Alibaba denied supplying H200s.

In September Alibaba released Qwen3.8-Omni-Flash, a native omni-modal model with a 1-million-token context window. Across 29 benchmarks, its average score rose more than 25% over Qwen3.5-Omni-Plus, while API pricing for audio input fell more than 98% and audio-video input more than 93%. PrismML separately released Ternary Bonsai 2 27B, a compressed model built on Qwen3.8 27B that retains 98.2% of the base model's benchmark scores at 5.9GB under an Apache 2.0 license. Alibaba Cloud backs the Creative Foundry and Industry Forum at AI Film Festival Japan, running October 31 to November 2, 2026. In October, China's Ministry of Industry and Information Technology asked Alibaba and ByteDance to report how many NVIDIA RTX Pro 5500 GPUs they want and what they will be used for, and plans to approve the purchases.

Key facts

Qwen3.8-Omni-Flash context window
1 million tokens ↗
Qwen3.8-Omni-Flash API price change
Audio input fell more than 98%; audio-video input more than 93% ↗
Qwen3.8 27B Intelligence Index
52, tying GPT-5.6 Luna ↗
Qwen3.8-Max parameters
2.4 trillion total, 95 billion active ↗
Qwen3.8-Max revenue sharing plan
Major commercial users will be required to share revenue; rate undecided ↗
Qwen-Audio-3.0-TTS ranking
Plus model ranks first on Artificial Analysis TTS leaderboard ↗
Qwen-Robot Suite models
Qwen-RobotNav, Qwen-RobotManip, Qwen-RobotWorld ↗
NVIDIA RTX Pro 5500 purchase status
China's MIIT asked Alibaba and ByteDance to report desired quantities and uses; plans to approve ↗
Moonshot AI chip contract
Approximately 20,000 NVIDIA chips via Alibaba, reportedly older Hopper architecture ↗
AI Film Festival Japan dates
October 31 to November 2, 2026, with Alibaba Cloud backing the Creative Foundry and Industry Forum ↗

Timeline

Synthesized by Yomimono from the cited Yomimono stories below, each itself sourced, then editorially reviewed. Every fact links the story it came from.

Connections

Structured graph also available as JSON at /public/entities/alibaba. CC BY 4.0.

All coverage

Oct 5

China Moves to Approve Alibaba and ByteDance Buys of NVIDIA RTX Pro 5500

China's Ministry of Industry and Information Technology has asked select domestic companies, including Alibaba and ByteDance, to report how many NVIDIA RTX Pro 5500 GPUs they want and what they will be used for, and plans to approve the purchases, The Information reports. The RTX Pro 5500 is a workstation GPU NVIDIA announced in September 2026 for large-scale AI models and simulations.

Sep 18

Alibaba Adds Qwen3.8-Omni-Flash To Qwen Model Line

Alibaba released Qwen3.8-Omni-Flash, a native omni-modal model in its Qwen line. It supports a 1-million-token context window and keeps text performance equal to text-only models of the same size. Across 29 benchmarks, its average score rose more than 25% over Qwen3.5-Omni-Plus, while API pricing for audio input fell more than 98% and audio-video input more than 93%.

Sep 18

PrismML Releases Ternary Bonsai 2 27B At A 5.9GB Footprint

PrismML announced Ternary Bonsai 2 27B, a compressed model built on Alibaba's Qwen3.8 27B. It ships at 5.9GB, one-ninth the size of the base model, and retains 98.2% of Qwen3.8 27B's benchmark scores. PrismML lists gains in reasoning, coding, vision, and long-horizon agent performance over the earlier Bonsai 27B. The model is released under the Apache 2.0 license and published on Hugging Face. PrismML says it runs on Mac, iPhone, and iPad.

Sep 15

AIFJ2026 Runs October 31 To November 2 With Alibaba Cloud, CapCut, ElevenLabs

AI Film Japan announced on September 7 that the second AI Film Festival Japan will run from Saturday, October 31 to Monday, November 2, 2026, centered on Tokyo Innovation Base. Alibaba Cloud backs the Creative Foundry and Industry Forum; CapCut and ElevenLabs each establish a sponsored special award inside the existing judging system. The first edition drew 563 works from more than 45 countries and regions.

Aug 12

Alibaba Plans Revenue Sharing for Big Users of Qwen3.8-Max

Alibaba will make Qwen3.8-Max an open model with open weights, but plans to require major commercial users to share revenue, according to sources. The move follows Moonshot AI's Kimi K3 license, which mandates commercial contracts for businesses with annual sales over $20 million. Alibaba's revenue sharing rate is undecided.

Aug 10

Alibaba Unveils Qwen3.8-Max With 2.4 Trillion Parameters

Alibaba's Qwen team announced Qwen3.8-Max, a multimodal AI with 2.4 trillion total parameters and 95 billion active parameters. It handles text, images, and video, with a 1 million token context. In some benchmarks it surpassed Anthropic's Claude Fable 5. The model ran 16 days autonomously, producing 265 commits and 127 pull requests.

Aug 9

Moonshot AI Used About 20,000 NVIDIA Chips Via Alibaba, Bloomberg Reports

Bloomberg reported that Moonshot AI signed a contract with Alibaba to use approximately 20,000 NVIDIA chips for its Kimi series AI models. The chips are reportedly older Hopper architecture, possibly H200, though Alibaba denied supplying H200s. The news follows Kimi K3's release, which scored higher than GPT-5.6 Sol and Claude Fable 5 in some benchmarks.

Aug 9

Alibaba's Qwen-Audio-3.0-TTS Tops Artificial Analysis Ranking

Alibaba's Tongyi Lab released Qwen-Audio-3.0-TTS, a text-to-speech model with two variants. The Plus model ranks first on Artificial Analysis's TTS leaderboard. It supports 16 languages including Japanese, offers natural-language control over emotion and pacing, and can clone voices from noisy reference audio.

Jun 17

Alibaba's Qwen Team Launches Qwen-Robot Suite for Robot Automation

Alibaba's Qwen AI research team announced Qwen-Robot Suite, a set of three AI models for robot development: Qwen-RobotNav for navigation and autonomous driving, Qwen-RobotManip for operating manipulators, and Qwen-RobotWorld for world recognition. The suite is designed to enable robots to follow natural language instructions and perform complex tasks.