PulseAugur
EN
LIVE 00:38:07

Tencent releases Hy3, a 295B parameter MoE model

By PulseAugur Editorial · [3 sources] ·

Tencent has released Hy3, a 295 billion parameter Mixture-of-Experts model developed by its Hy Team. This model, featuring 21 billion active parameters, has demonstrated superior performance compared to similarly sized models and rivals larger open-source alternatives. Hy3 offers a context length of 256K and is available for free on OpenRouter until July 21st. AI

IMPACT This release offers a powerful new open-source option for developers, potentially driving innovation in agentic workflows and productivity tasks.

RANK_REASON Frontier-lab model release with system card. [lever_c_demoted from frontier_release: ic=2 ai=1.0]

Read on Hugging Face Trending Models →

AI-generated summary · Google Gemini · from 3 sources. How we write summaries →

Tencent releases Hy3, a 295B parameter MoE model

COVERAGE [3]

  1. Simon Willison TIER_1 English(EN) ·

    tencent/Hy3

    <p><strong><a href="https://huggingface.co/tencent/Hy3">tencent/Hy3</a></strong></p> New Apache 2.0 licensed model from Tencent in China:</p> <blockquote> <p>Hy3 is a 295B-parameter Mixture-of-Experts (MoE) model with 21B active parameters and 3.8B MTP layer parameters, developed…

  2. Hugging Face Trending Models TIER_1 English(EN) · tencent ·

    tencent/Hy3-FP8

    text-generation · 443 downloads · 49 likes

  3. Hugging Face Trending Models TIER_1 English(EN) · tencent ·

    tencent/Hy3

    text-generation · 2 downloads · 181 likes

RELATED ENTITIES

RELATED TOPICS