Concept

Mixture of Experts

A model design that routes each input to a subset of specialised sub-networks for efficiency.

A full explainer for this concept is being written. In the meantime, here's what's in the news.

In the news

Inside the Model Factory — Eiso Kant, Poolside AI

Latent Space · Jul 23, 2026

Latent-MoE — Implementation of LatentMoE,Toward Optimal Accuracy per FLOP and Parameter in Mixture of Experts (Elango et al., NVIDIA

GitHub · Jul 21, 2026

China’s Low-Priced Z.ai Model Is Exposing Costly Coder Habits

IEEE Spectrum · Jul 21, 2026

Inkling: Our open-weights model

Simon Willison · Jul 16, 2026

pulsar — SSD-streaming inference engine for giant MoE models (Rust + CUDA). GLM 5.2 743B at 2 tok/s and Hy3 295B at 7 tok/s on tw

GitHub · Jul 13, 2026

Product Hunt · Jul 7, 2026

Simon Willison · Jul 6, 2026

Tencent releases Hy3 open-source model that allegedly matches models up to five times its active size

The Decoder · Jul 6, 2026