State of Open Models: Summer 2026 Observations
Hugging Face's mid-2026 analysis reveals a shifting open-model landscape: Chinese labs now lead the frontier with the largest models, while U.S. hardware vendors like AMD and NVIDIA dominate new releases. Qwen has become the community's foundational model, and small models under 1B parameters account for 83% of all-time downloads, with llama.cpp enabling trillion-parameter models to run locally.
Hugging Face
Moonshot AI
MiniMax
Tencent
Alibaba/Qwen
Meituan
NVIDIA
Liquid AI
Hugging Face blog14.08 · 20:04
