CoinWorld News, Tencent has released the latest flagship model of the Hunyuan series, Hy3 preview, adopting a MoE (Mixture of Experts, where multiple small expert networks are activated on demand) architecture, with a total of 295 billion parameters and 21 billion activated parameters. The model has been launched on the OpenRouter API and Hugging Face demo, but the open-source weights have not yet been released. This is the first flagship model led by former OpenAI researcher and Tencent Chief AI Scientist Yao Shunyu after he joined, and it is also the output of Tencent's rebuilding of its foundational model R&D system in February of this year, produced in less than three months. Tencent positions it as "on par with top domestic models."

View Original
This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.
  • Reward
  • 6
  • 2
  • Share
Comment
Add a comment
Add a comment
YieldFarmLibrarian
· 3h ago
Tencent moved pretty fast this time, rebuilding the system in February and launching a flagship in May. Yao Shunyu's team leadership efficiency is impressive.
View OriginalReply0
StardustUnderTheGlassDome
· 5h ago
OpenRouter and Hugging Face are both on board, but the open-source weights haven't been released yet, just teasing, right?
View OriginalReply0
ShortPositionsAtTheElevator
· 5h ago
Former OpenAI employee becomes Chief Scientist, Tencent AI has big ambitions
View OriginalReply0
WhirlpoolInATeacup
· 5h ago
Domestic top-tier models on par? Modest, let's see the actual measurements.
View OriginalReply0
BullsAndBearsInVinyl
· 5h ago
If the open-source weights are not available, wait and see. What real ability can an API demo test?
View OriginalReply0
RetroRadioIridescence
· 5h ago
MoE architecture has become the standard now, the key lies in training data and post-processing.
View OriginalReply0
  • Pinned