Tencent Hy3 Review: 21B-Active Open MoE That Beat GLM-5.1 in a Blind Test

2026-07-12 · Source: LLM on Medium · Field: Technology & Digital — Artificial Intelligence & Machine Learning · Depth: Intermediate, quick

Summary

Tencent launched Hy3 on July 6, 2026, an open Mixture-of-Experts (MoE) model featuring 295 billion total parameters with 21 billion active parameters and a 256K context window. Released under an Apache 2.0 license, Hy3 is available on OpenRouter and Tencent Cloud, with free inference offered until July 21, 2026. Notably, Tencent subjected Hy3 to a public blind head-to-head comparison against Zhipu's GLM-5.1, where 270 domain experts graded both models on real-world tasks. Hy3 emerged victorious, scoring 2.67 compared to GLM-5.1's 2.51 across 312 valid comparisons on a four-point scale. This public benchmarking signifies a strategic shift for Tencent, moving beyond internal tooling to establish itself as a frontier lab in the open AI ecosystem.

Key takeaway

For Machine Learning Engineers evaluating open-source MoE models, Hy3's performance in a blind test against GLM-5.1 suggests it warrants serious consideration. You should benchmark Hy3's 21 billion active parameters and 256K context for your specific routing needs, especially given its Apache 2.0 license and free inference availability on OpenRouter and Tencent Cloud until July 21, 2026. This shift by Tencent indicates a new competitive landscape for high-performance open models.

Key insights

Tencent's Hy3 MoE model won a public blind test against GLM-5.1, signaling Tencent's strategic pivot to open frontier AI development.

Principles

• Public benchmarking validates model performance.
• Open source models can challenge incumbents.
• Strategic shifts require public demonstration.

Method

The article describes a blind head-to-head test where 270 domain experts graded Hy3 and GLM-5.1 on real work, using a four-point scale across 312 comparisons.

In practice

• Evaluate Hy3 on OpenRouter or Tencent Cloud.
• Compare Hy3's 21B active parameters.
• Utilize free inference until July 21, 2026.

Topics

• Tencent Hy3
• Mixture-of-Experts
• Large Language Models
• Open-Source AI
• Model Benchmarking
• Tencent Cloud

Best for: CTO, VP of Engineering/Data, AI Engineer, AI Scientist, Machine Learning Engineer, Director of AI/ML

Related on AIssential

• tencent/Hy3 — Tencent's Hy3 MoE model delivers competitive performance against much larger models while maintaining a more efficient active parameter count.
• Tencent HY3 IS REALLY GOOD! Best Open-Weight Model? (FULLY FREE) — Tencent's HY3 is a 295B MoE model offering competitive performance against larger models at a lower cost, especially for coding and creative tasks.
• GLM-5.2 Is The New Best Open Model — GLM-5.2 sets a new benchmark for open models, but its Claude distillation implies generalization limits and a "benchmaxxed" performance profile.
• NEW Tencent Hy3 is here for FREE! — Tencent Hy3, an efficient MoE LLM, excels in agentic coding with high performance and strong benchmarks against larger models.
• Tencent's Apache-licensed Hy3 takes on GLM-5.2 at half the size — and wins everywhere except coding — Tencent's Hy3 offers a reliable, deployable open-weight MoE model with a permissive license, balancing performance with resource efficiency.
• The New Frontier: Benchmarking China’s Open-Weight Triad GLM 5.1, KIMI 2.6, MiMo-V2.5-Pro — Open-weight Chinese MoE models are outperforming proprietary Western AI on key developer benchmarks.

See Counsel's argued verdicts on the open AI decisions leaders are weighing →

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by LLM on Medium.