Comment by anana_

For more context, this puts it on par with models like GLM 5.2 and GPT 5.6 Luna, which are far larger