Alibaba Previews Qwen3.8-Max, a 2.4 Trillion-Parameter Multimodal Model, Days After Moonshot’s Kimi K3 Open-Weight Launch
Alibaba's Qwen team has previewed Qwen3.8-Max, a 2.4 trillion-parameter multimodal model. The model is currently available for preview at a reduced price.
- Alibaba's Qwen team has previewed Qwen3.8-Max, a 2.4 trillion-parameter multimodal model
- The model is available for preview at a reduced price of 10% of the standard pricing
- Key details such as benchmark tables and model cards are not yet available
- The model is designed to handle multiple types of data, including text, images, and audio
Alibaba's Qwen team has introduced Qwen3.8-Max, a massive multimodal model boasting 2.4 trillion parameters. This model is being touted as the second most powerful, after Fable 5. The preview of Qwen3.8-Max is currently live on several platforms, including Token Plan, Qoder, and QoderWork, at a discounted price of 10% of the standard pricing.
The launch of Qwen3.8-Max comes on the heels of Moonshot's Kimi K3 Open-Weight launch, highlighting the increasing competition in the AI model development space. However, it is worth noting that key details such as benchmark tables, model cards, licenses, per-token prices, and active-parameter counts are not yet available.
The Qwen3.8-Max model is a multimodal MoE model, which means it is designed to handle multiple types of data, such as text, images, and audio. The model's massive parameter count suggests that it has the potential to achieve state-of-the-art results in various AI tasks.
The preview of Qwen3.8-Max is an opportunity for developers and researchers to explore the capabilities of this powerful model and understand its potential applications. However, the lack of detailed information about the model's performance and pricing may limit its adoption in the short term.
As the AI landscape continues to evolve, the development of large-scale multimodal models like Qwen3.8-Max is likely to play a significant role in shaping the future of AI research and applications.
offers a powerful tool for developing AI applications
has the potential to drive innovation and improve efficiency
represents a significant investment in AI research and development
advances the state-of-the-art in AI research
- MoE
- Mixture of Experts, a type of neural network architecture
- multimodal
- capable of handling multiple types of data, such as text, images, and audio
LLMSomeone Fine-Tuned OpenBMB’s MiniCPM5-1B on Claude Fable 5 Traces to Ship a 657MB Local Thinking Model
Alibaba previews Qwen3.8, claims it’s second only to Claude Fable 5 - SiliconANGLE
LLMGPT-5.6 Sol yields 30-year math proof as METR flags severe evasion behaviors
LLMKimi: Threat or menace?
Chinese AI Model Takes U.S. by Surprise with Abilities Rivaling Claude and ChatGPT - Broadband Breakfast
AI ToolsAI agents, meet the Azure Cosmos DB vNext emulator
Azure Cosmos DB vNext emulator now supports AI agents for local development, improving efficiency and accuracy.
Head of US AI safety agency resigns - Reuters
The head of the US AI safety agency has resigned, sparking concerns about the future of AI regulation in the country.
AI in 2026: Smarter Models, Harder Questions - USC Viterbi School of Engineering
The USC Viterbi School of Engineering discusses the future of AI in 2026, highlighting smarter models and harder questions. Experts predict significant advancements in AI technology.
AI ToolsIntroducing Cosmos 3 Edge
Nvidia has introduced Cosmos 3 Edge, a new AI model designed for edge computing applications. It's built on the company's existing Cosmos 3 architecture.
AI is writing, acting and producing China’s minidramas. It’s shaking a $14 billion industry. - NBC News
AI is being used to write, act, and produce minidramas in China, shaking a $14 billion industry.
AI ResearchAt SIGGRAPH, NVIDIA Advances Graphics and Simulation With Agentic and Physical AI
At SIGGRAPH 2026, NVIDIA announced new agentic and physical AI technologies that enhance real‑time graphics rendering and simulation, aiming to improve media creation and robotics.