Tencent launches Hy4 flash, cutting AI task costs by 95% in internal tests
technode.com
technode.com
https://www.tencent.com/tencent-launches-hy4-flash-to-power-...
I've been using deepseek 1017 since it came out (and its little brother Particle-1) and I've been extremely satisfied. Best cost to intelligence model available.