Ling-3.0-flash-Fin went live on a Western routing layer at 15:58 UTC on 27 August, hours before the Chinese-language announcement. Ant says the weights come "next week"; its Hugging Face organisation currently lists no such repository.
2d ago
The lab's first in-house model is a sparse mixture-of-experts system trained on 45 trillion tokens and released with full weights — and the company openly concedes it is not the strongest model available, open or closed.
Jul 16, 2026
Hunyuan 3.0 is a 295-billion-parameter mixture-of-experts model that fires just 21B parameters per token, ships under Apache 2.0, and edges GPT-5.5 on science — free on OpenRouter through July 21.
Jul 8, 2026
