rotative
huggingface blog · November 18, 2026

Hugging Face pushes MoE inference limits with 180B param model

hugging face releases qwen3.8-flash-next-gsq-rco-gguf with 177b parameters, impacting large model deployment.

read in 简体中文 · 繁體中文

hugging face has released a new model, qwen3.8-flash-next-gsq-rco-gguf, with 177 billion parameters. this image-text-to-text model was updated on september 29 and has already garnered significant attention with 7.3 million downloads and 1.31k views. the release of this large model pushes the limits of mixture of experts (moe) inference. the deployment of such large models has implications for the development and application of artificial intelligence.

read at source ↗

← rotative