AI BriefingTencent Unveils 295B MoE Model Hy3
·2026.07.07 08:57
Key point
Tencent has released Hy3, a 295B-parameter MoE model, under the Apache 2.0 license.
Details
Developed by the Tencent Hy Team, Hy3 is a Mixture-of-Experts (MoE) model with a total of 295B parameters, using 21B active parameters and 3.8B MTP layer parameters at inference time.
Key features are as follows:
- Performance: Outperforms similarly sized models and shows performance competitive with flagship open source models that have 2 to 5 times more parameters.
- Context Length: Supports 256K.
- License: Released under the Apache 2.0 license, allowing free use.
- Model Size: The full model is 598GB, and the FP8 quantized version is 300GB.
It is currently available for free testing via OpenRouter until July 21.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
Tencent Releases 'Hy4 Preview', a 770B Parameter Open-Weight LLM
Step 5 Preview Released: 600B Parameter MoE Model Highlights Agentic Tasks and Financial Analysis Performance
Xiaomi Releases Open-Weight Multimodal Model MiMo-V2.6-Pro… Higher Intelligence and Competitive Pricing Compared to Similar-Sized Models