Bullish
Alibaba Qwen3.8-2.4T-A95B Model Launches on Supercomputing Internet
21:23
Alibaba open-sources Qwen3.8-2.4T-A95B MoE model on supercomputing internet, enabling direct deployment via FlagOS-optimized images for heterogeneous hardware.
Woofun AI reports that Alibaba has released its newly open-sourced ultra-large-scale Mixture of Experts model, Qwen3.8-2.4T-A95B, on the supercomputing internet. Leveraging a massive supercluster with thousands of GPUs at core nodes, developers can directly utilize Qwen3.8-2.4T-A95B model images within the AI community.
These images are optimized for a unified software stack based on FlagOS and domestic heterogeneous acceleration cards, allowing businesses to rapidly deploy AI application services tailored to specific operational requirements.
WOOFUN AI
Impact Assessment · Quick Read
The release of Qwen3.8-2.4T-A95B on the supercomputing internet lowers deployment barriers for large-scale MoE models by providing pre-optimized images for domestic hardware. This move may accelerate adoption among enterprises seeking sovereign AI infrastructure solutions compatible with Chinese acceleration cards. The availability of such resources could intensify competition in the domestic LLM deployment sector.
Generated by WOOFUN AI · For reference only, not investment advice
Comments
No comments yet.