2026-10-11 17:20 UTC

Tencent Hunyuan claims Hy4 Preview is a usable open LLM release for local deployment, potentially expanding the range of independently inspectable and self-hostable models.

state: expiredheat: lowuncertainty: highconvergesscott: lowopen-models local-inferenceTencent Hunyuan

What is this?

Tencent Hunyuan announced Hy4 Preview as an open-sourced mixture-of-experts language model with 770B total parameters, 49B active parameters, and a context window exceeding 1M tokens. Tencent positions it for productivity tasks including coding, office work, and scientific research, with distribution through GitHub, Hugging Face, ModelScope, Tencent Cloud, and OpenRouter. The supplied snippets establish inspectable/open distribution but do not specify the license, hardware requirements, independent evaluations, or evidence that deployment on ordinary local hardware is practical.

Why it matters to Scott

Tencent’s inspectable, distributable release points toward Scott’s model-swappability and sovereign-software position, but currently only as another candidate model rather than demonstrated independent operability. Without license clarity, hardware requirements, or independent evaluations showing that a 770B/49B-active MoE is practical to self-host, it does not yet change his local-inference stack or claims.
ip:framework.sovereign-software-assuranceip:concept.model-perishabilityip:concept.capability-auditdev:project.gamepcdev:concept.hardware-aware-local-inferenceradar:concept.open-modelsradar:concept.local-inferenceradar:concept.mixture-of-expertsradar:concept.model-evaluation
queries asked of Scott's wikis
  • open weights versus open source model criteria
  • local inference hardware and economics
  • self-hosted model sovereignty strategy
  • mixture-of-experts deployment constraints
  • independent model evaluation and inspectability
  • local models for coding-agent infrastructure

Measured heat

no measured readings yet — the hourly heat pass fills this in

How the heat travelled

no chain yet — the hourly chain pass fills this in

Evidence (5) — ⭐ canonical anchor

sourceobjectauthorscorecomments
🟧 hnTencent Hy4 Preview LLMitvision20
🟧 echo.github ⭐The repository presents Tencent's Hy4 Preview LLM release.Tencent Hunyuan——
🟠 redditAn official 1-bit quant for Hy4??? 👀
LocalLLaMA
lakySK619
🟠 redditTencent compressed Hy4-preview from 1.5TB to about 200GB GGUF and kept about 98% performance.
LocalLLaMA
RedditUsr2874140
🟧 hnTencent Releases and Open-Sources Tencent Hy4 Previewshenli3514381232

Interpretation history

Decision trace