/
Navigation
Chronicles
Browse all articles
Explore
Semantic exploration
Research
Entity momentum
Nexus
Correlations & relationships
Story Arc
Topic evolution
Drift Map
Semantic trajectory animation
Posts
Analysis & commentary
Pulse API
Tech news intelligence API
Browse
Entities
Companies, people, products, technologies
Domains
Browse by publication source
Handles
Browse by social media handle
Detection
Concept Search
Semantic similarity search
High Impact Stories
Top coverage by position
Sentiment Analysis
Positive/negative coverage
Anomaly Detection
Unusual coverage patterns
Analysis
Rivalry Report
Compare two entities head-to-head
Semantic Pivots
Narrative discontinuities
Crisis Response
Event recovery patterns
Connected
Search: /
Command: ⌘K
Embeddings: large
TEXXR

Chronicles

The story behind the story

days · browse · Enter similar · o open

Tencent releases Hy3, a 295B-parameter model that it says is competitive with GLM-5.1 and GLM-5.2, under the Apache 2.0 license, after a preview launch in April

For the past year, the awkward secret of the open-weight model boom has been that many of the strongest Chinese releases …

VentureBeat Sam Witteveen

Context & Ripple Effects

Tencent’s April Hy3 preview marked its first model developed under former OpenAI researcher Yao Shunyu and reduced the stated parameter count from HY2’s 400B to 295B. This release turns that preview into an Apache 2.0-licensed model offering.

The launch lands amid a rapid sequence of open-weight releases from Z.ai, whose GLM-5.1 is also MIT-licensed and presented as a high-end benchmark competitor. Tencent is positioning Hy3 against that specific open-model reference point rather than introducing an isolated research project.

First-order effects

  • Developers and enterprises can now use, modify, and redistribute Hy3 under Apache 2.0 terms, lowering licensing friction for teams evaluating a large Tencent model.
  • Tencent gains a public, directly comparable entrant in the high-end open-weight segment; its performance position remains based on the company’s claim of competitiveness with GLM-5.1 and GLM-5.2.

Second-order effects

  • Z.ai and other open-model providers face a more crowded comparison set for customers choosing among permissively licensed Chinese models, increasing the importance of reproducible evaluations and deployment economics rather than parameter counts alone.
  • The use of Apache 2.0 alongside Z.ai’s MIT licensing makes license permissiveness less of a differentiator between these offerings, shifting evaluation toward model capability and practical integration.

Third-order effects

  • If leading Chinese developers continue publishing large models under permissive licenses, enterprise AI stacks may become less tied to a single proprietary model provider and more dependent on interchangeable model evaluation and deployment tooling.
  • The competitive center of gravity may move from simply releasing open weights to demonstrating dependable performance on real tasks; vendor benchmark claims will need independent validation to determine whether that shift is durable.

The trend: Hy3 is part of an intensifying race among Chinese AI developers to pair frontier-scale model claims with permissive open-weight licenses.

Discussion

  • @tencenthunyuan Tencent Hy on x
    🚀Hy3 is here. 295B MoE. Best in its size class. Rivals trillion-scale flagships. Reliable and affordable for most agentic usecases. Apache 2.0. Friendly for commercial use. FREE API for 2 weeks → https://openrouter.ai/... 🤗 https://huggingface.co/... 📖 https://hy.tencent.com/... …
  • @shunyuyao12 Shunyu Yao on x
    Hy2 -> Hy3 preview -> Hy3 Another massive leap forward, under half a year. Not just a leap of reasoning or agentic capabilities. Also a leap of anti-hallucination, reliability, and product experiences. More on the way and so proud of the team! 🧑‍🍳🧑‍🍳🧑‍🍳 [image]
  • @jenzhuscott Jen Zhu on x
    So @TencentHunyuan @ShunyuYao12 did not disappoint - just dropped Hy3. Apache 2.0. The numbers 🔥🔥 - 295B total params, 21B active. Compare to DeepSeek/Qwen at ~1T, overseas frontier at ~10T. - API: ¥1 / ¥4 / ¥0.25 per M tokens. Cheapest Chinese model on the market. Cheaper [image…
  • @rogerw0108 Roger Wang on x
    🔥Huge congrats to the @TencentHunyuan team for the incredible model release! Alongside Hy3, we've been working closely with the team to bring their HPC-Ops natively into vLLM! See full technical details in our blog post and take Hy3 for a spin today🚀 https://vllm.ai/...
  • @gneubig Graham Neubig on x
    Interesting new release Hy3. If the numbers are really reflected in real-world vibes, this is a step closer to a strong model you can host yourself (more easily than GLM-5.2).
  • @tencentai_news @tencentai_news on x
    Introducing Hy3, our latest Hy model🙌 It outperforms similar-sized models and rivals flagship open-source models with 2 to 5× the parameters. — Strong, practical gains in coding, office productivity, financial modeling, frontend design, and game development — Blind-tested by
  • @zixuanli_ Zixuan Li on x
    Hy3 from @TencentHunyuan is out. Great to see real-world workflows emphasized, with GLM-5.1 cited in the comparison: “we ran a blind test with 270 experts from various disciplines, working on real-world workflows. Hy3 scored 2.67/4, outperforming GLM-5.1 at 2.51/4.” [image]
  • @tencenthunyuan Tencent Hy on x
    Hy3 is a real productivity booster. 🚀See how it handles these two workflows. [video]
  • @atomic_chat_hq @atomic_chat_hq on x
    New Hunyuan Hy3 hits Gemini 3.5 quality on physics for 35x cheaper! We gave 4 models the same prompt: build three self-contained HTML5 canvas scenes with real physics demos Prompts: - A bowling ball knocking down the pins - An air hockey rally that ends in a goal - A pool [video]
  • @teksedge David Hendrickson on x
    🚀 Hy3 is officially here from Tencent Hunyuan! Full Details 💡 What Makes This Different? 🔹 DeepSeek V4 class intelligence, probably runnable on 128GB RAM (maybe Q4) 🔹 Hy3 achieves 74.4% on SWE-Bench Verified 🔹 Reduces hallucinations (down to ~5.4% in evaluations) Model [image]
  • @sentdex Harrison Kinsley on x
    Oh my this looks really good. We're being inundated with increasingly exceptional models in OSSAI. No quotas your model your data model behavior doesn't change
  • @hrudolph Hannes Rudolph on x
    OpenClaw FREE MODEL alert!! Hy3 just released from @TencentHunyuan and you can use it in @openclaw now FREE through @OpenRouter for 2 weeks! Run this to configure: openclaw models set openrouter/tencent/hy3:free [image]
  • @zhengyang_42 Zhengyang Tang on x
    3 months of focused team iteration wrapped up — Hy3 is officially live. This 295B MoE model delivers flagship-level performance matching trillion-scale alternatives, optimized for stable, accessible agent deployments. Licensed under Apache 2.0 for unrestricted commercial use,