{
  "id": 4309122,
  "title": "Introducing Hy4 Preview",
  "url": "https://urgent.news/2026/08/29/introducing-hy4-preview",
  "topic": "ai",
  "section": "AI",
  "published": "2026-08-29T23:53:13.000Z",
  "source": {
    "name": "Simon Willison",
    "slug": "simon-willison",
    "url": "https://simonwillison.net/2026/Aug/29/hy4/"
  },
  "original_language": "en",
  "account": "Introducing Hy4 Preview. Tencent, a Chinese tech company, has unveiled its latest open weight text input language model (LLM) named Hy4. This new model boasts an impressive 770 billion total parameters, with 49 billion active parameters and a context window of one million tokens. The model's files weigh in at 1.56 terabytes, which are available on Hugging Face. This represents a substantial jump from their previous model Hy3, released in July, which had 295 billion total parameters, 21 billion active parameters, a context window of 256,000 tokens, and a file size of 598 gigabytes.\n\nCurious about Hy4's capabilities, the author experimented with model chat templates to better understand its potential. They discovered that Hy4 offers only two reasoning effort levels: the default \"high\" setting and the \"no_think\" option, which disables reasoning altogether. The author put this feature to the test by prompting Hy4 to generate an SVG of a pelican riding a bicycle with high reasoning enabled. The result, quoted from the prompt, reads: \"...Let's maybe add a helmet? It could improve riding theme, but may obscure head. Maybe a small cycling cap or helmet? The user didn't ask; can add red helmet? Might be cute. But pelican with big beak; a helmet might obscure. Better maybe no. Maybe add sunglasses? No. Maybe add water? No.\"\n\nThe text acknowledges the reasoning trace's use of truncated English, likely due to grammar not being essential for hidden reasoning purposes. The article concludes with a link to a post by Simon Willison, dated 29th August 2026, inviting readers to sponsor the author for $10 per month in exchange for a curated email digest of the month's most important LLM updates.",
  "summary": "Introducing Hy4 Preview New open weight text input (no vision) LLM from Chinese company Tencent today: 770B total parameters, 49B active parameters, 1M token context window, 1.56TB on Hugging Face . This is a big size increase from their previous Hy3 in July, which was 295B, 21B active, 256,000 context, 598GB. I recently started using model chat templates to better understand their capabilities.…",
  "key_points": [
    "Tencent unveils Hy4, a 770 billion parameter LLM with 49 billion active parameters.",
    "Context window of Hy4 extends to one million tokens, a significant increase from Hy3.",
    "Hy4 offers only two reasoning effort levels: high and nothink."
  ],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}