Intel / post

Hy3 preview opens at 295B with 21B active, aimed at agents

Source: x.com
Date: Published Apr 23, 2026

Context

Tencent's Hy3 preview is a downloadable designed for reasoning, coding and software- tasks. Its model card explains the 295B-A21B shorthand: the model contains 295 billion parameters, with 21 billion active for each text . Its design selects eight of 192 expert components at each relevant processing step.

Selective activation limits how much of the model performs calculations for each token, while the complete model remains much larger than the active portion. The preview therefore offers developers access to a large model for experimentation; the active-parameter figure alone does not describe the hardware needed to host it.

Terms in this piece · Glossary

LLM — A large language model — the neural network behind tools like Claude and ChatGPT, trained on huge amounts of text to predict what comes next.
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
mixture-of-experts — A model built from many specialist sub-networks where only a few activate per token, giving big-model capability at small-model running cost.

Read the source x.com