Tencent’s Hunyuan team released and open sourced the Hy4 preview large language model on August 28, 2026 and global tech outlets published detailed breakdowns on August 30, 2026. Hy4 uses a 770 billion parameter mixture of experts architecture with 49 billion active parameters and a context window of over 1 million tokens, with weights released under an Apache 2.0 style license.
This article aggregates reporting from 4 news sources. The TL;DR is AI-generated from original reporting. Race to AGI's analysis provides editorial context on implications for AGI development.
Hy4 preview matters because it pushes open weights back into a race that has been tilting toward closed frontier labs. A 770 billion parameter mixture of experts model with a million token context window and an Apache style license gives independent teams and national ecosystems a serious tool to experiment with long context software engineering, document heavy workflows and scientific computing. That combination of scale, permissive licensing and claimed production readiness is rare, and it puts pressure on rivals to justify why their best models remain locked behind API terms.
What makes Hy4 more interesting than its raw size is the way Tencent is marketing it. The company is framing Hy4 as an “engineering brain” for long running, multi step work rather than a general purpose chatbot. Internal evaluations focus on real coding and production tasks, and official demos show the model orchestrating other models to explore research directions. That hints at a broader shift where top vendors sell not just raw intelligence but model centric workflows that coordinate fleets of tools. For the race to AGI, Hy4 reinforces a pattern where open models keep catching up on coding and reasoning, while closed players still lead on absolute performance and safety research depth. The gap is smaller than it was a year ago, and that dynamic will influence how regulators and enterprises think about concentration of power in a handful of US labs.