Tencent Open-Sources Hy3: 295B MoE Model Rivals Larger Flagships

yshan2u · x · 2026-07-06

Tencent has open-sourced the Hy3 model on Hugging Face. Featuring an MoE architecture, it has 295B total parameters and roughly 21B activated parameters. Reportedly, its reasoning and coding capabilities rival flagship models 2 to 5 times its size, achieving high performance with a smaller activated compute footprint, which has sparked widespread discussion in the community.

Related event: Tencent Open-Sources Hunyuan Hy3 for Agentic Workloads(26 posts)→

Original post →

More from Models

Models channel →