Hy3-preview

Name: Hy3-preview
Author: Tencent

Tencent🇨🇳 China

active

Compare with other models →

Context window256K tokens

Version History

previewsnapshotApril 23, 2026

First model trained on Tencent's rebuilt infrastructure. Emphasizes complex reasoning, instruction following, context learning, coding, and agent tasks with rebuilt RL infrastructure.

Benchmark Scores

Full leaderboard →

76.3%

MATH

87.4%

MMLU

65.8%

MMLU-Pro

Coverage

model releaseTencent

Tencent Releases Hy3-Preview: 295B-Parameter MoE Model with 21B Active Parameters

Tencent has released Hy3-preview, a 295-billion-parameter Mixture-of-Experts model with 21 billion active parameters and a 256K context window. The model scores 76.28% on MATH and 34.86% on LiveCodeBench-v6, with particularly strong performance on coding agent tasks.

April 23, 2026 · 8:51 PM2 min read

Tencent Hy3-preview MoE