Why is nobody talking about Tencent’s Hy3 Preview?
Summary
The article observes that Tencent's Hy3 Preview open model performs surprisingly well in evaluations, narrowing the gap with top closed models, yet remains underdiscussed compared to Western AI labs.
Similar Articles
HY-3 PREVIEW
Tencent releases Hy3-preview, a 295B-parameter MoE model with 21B active parameters that excels in STEM reasoning, instruction following, coding and agent tasks.
The mysterious Hy3 LLM is topping OpenRouter Model Rankings by a large margin
A mysterious model called Hy3 from Tencent has unexpectedly topped OpenRouter's LLM rankings by token usage, despite mediocre benchmark performance and low public awareness. The article investigates the anomaly using OpenRouter's public data.
How far behind are open models? (17 minute read)
An analysis from LessWrong examining the performance gap between open-source and proprietary AI models.
Claude Mythos, Deepseek v4, HappyHorse, Meta’s new AI, realtime video games: AI NEWS
Anthropic unveils a withheld Claude Mythos model that autonomously finds thousands of 0-days, ZAI open-sources the 1.5 TB GLM-5.1 that tops open-weight benchmarks, Alibaba’s unreleased HappyHorse video model hits #1 on public leaderboards, and Deepseek teases an “Expert Mode” v4 preview.
Qwen 3.6 Max Preview just went live on the Qwen Chat website. It currently has the highest AA-Intelligence Index score among Chinese models (52) (Will it be open source?)
Qwen 3.6 Max Preview launched on Qwen Chat website, achieving the highest AA-Intelligence Index score (52) among Chinese models, with uncertainty about whether it will be open source.