On August 5, Ant Bering officially opened the new generation of native hybrid inference model Ling-3.0-flash, focusing on extreme intelligent density, closed-loop full-link agent, and low-cost large-scale implementation. Shengteng simultaneously completed 0 day adaptation. In the process of adapting to Bering Ling-3.0-Flash, Shengteng introduced the CANN PyPTO operator programming framework for the first time, drastically shortened the delivery cycle of complex fusion operators, efficiently completed operator access and inference framework performance tuning, simultaneously opened up complete deployment projects, provided developers, government and enterprise industries with integrated implementation solutions with independent innovation and high performance, and continued to expand the domestic large-scale model ecosystem.

Zhitongcaijing · 1d ago
On August 5, Ant Bering officially opened the new generation of native hybrid inference model Ling-3.0-flash, focusing on extreme intelligent density, closed-loop full-link agent, and low-cost large-scale implementation. Shengteng simultaneously completed 0 day adaptation. In the process of adapting to Bering Ling-3.0-Flash, Shengteng introduced the CANN PyPTO operator programming framework for the first time, drastically shortened the delivery cycle of complex fusion operators, efficiently completed operator access and inference framework performance tuning, simultaneously opened up complete deployment projects, provided developers, government and enterprise industries with integrated implementation solutions with independent innovation and high performance, and continued to expand the domestic large-scale model ecosystem.