Alibaba Qwen3.8-Max narrows open-weight frontier gap
Qwen3.8-Max is Alibaba's 2.4T open-weight model ranking fifth in text and second in vision on Arena. The gap with US frontier labs narrows but does not close.
4 stories tagged mixture-of-experts.
Qwen3.8-Max is Alibaba's 2.4T open-weight model ranking fifth in text and second in vision on Arena. The gap with US frontier labs narrows but does not close.
Moonshot AI's open-weight Kimi K3 ranks fourth of 580 models on Artificial Analysis while activating only 104B of 2.8 trillion parameters.
Thinking Machines Lab's first open-weights model, Inkling, has 975B parameters, 41B active and an Apache 2.0 license, and targets fine-tuning over benchmarks.
LongCat-2.0 is a 1.6 trillion parameter AI model from Meituan that activates only 48 billion parameters per token. Its weights are now open under the MIT license.